Category Archives: sadmin

sadmin

Piwik: "Fatal error: Allowed memory size"

В един момент забелязвам, че тоя Piwik не брой и тъкмо щях да го ругая, че е поредната боза, реших да го разгледам по- отблизо. Видях, че колектора които се изпълнява "WEBDIR/misc/cron/archive.sh" казва, че php-то което се изпълнява няма памет (ей това си е сериозна причина да кажа, че е пълна шит , но както и да е(да го напишат на perl)). Съответно когато излезе тази грешка излиза и отчита посещенията само на сайтовете преди тази грешка За да не променям глобалния php conf само заради тоя piwik едитнах колектора или по- точно на 45 ред добавих следното:

PHP_BIN="$PHP_BIN -d memory_limit=1300M";

Идеята е когато изпълнява php-то да добави този аргумент за да ползва повече памет(но докога ще бъде така …)

Когато се пусне отново ще колектне и архивира всички данни за всички сайтове(освен ако и тази памет не му стига :( ).

VN:F [1.9.22_1171]
Rating: 5.0/5 (1 vote cast)
VN:F [1.9.22_1171]
Rating: 0 (from 0 votes)

MySQL малък трик

Един малък трик за експорт на таблици от база с конкретен префикс. Понеже mysqldump не позволява това ползвам ето този трик:

mysqldump -uUSER -pPASS TABLE $(mysql -uUSER -pPASS -D TABLE -Bse "show tables like 'PREFIX_%'") > /tmp/DATABASE.sql

VN:F [1.9.22_1171]
Rating: 5.0/5 (1 vote cast)
VN:F [1.9.22_1171]
Rating: 0 (from 0 votes)

Upgrade на Pligg от терминал

Upgrade на Pligg от terminal:

bf:# cat pligg_upgrader_executor.sh

#!/bin/bash
# DH Pligg Upgrader v 1.0 (Written by amri)

if [ "$#" -ne 2 ];
then
echo "Use arguments: install.sh PATH DOMAIN";exit;
fi;

PATH=$1;
DOMAIN=$2;
CP="/bin/cp";
UNZIP="/usr/bin/unzip";
RM="/bin/rm";
CURL="/usr/bin/curl";
CHMOD="/bin/chmod";
MV="/bin/mv";
WGET="/usr/bin/wget";

if [ ! -f "/tmp/latest.zip" ]
then
$WGET -q "http://pligg.com/downloads/latest/latest.zip" -O "/tmp/latest.zip";
fi;

if [ ! -d "$PATH" ]
then
echo "DIRECTORY $PATH NOT EXIST!";exit;
fi;

cd $PATH;
$CP /tmp/latest.zip . > /dev/null
$CP -Rvp templates templates.bak > /dev/null
$CP settings.php settings.php.bak > /dev/null
$CHMOD -R 777 modules languages cache settings.php > /dev/null
$UNZIP -o latest.zip > /dev/null
$CP -Rvp  templates.bak/* templates/ > /dev/null
$RM -fr templates.bak > /dev/null
$CHMOD -R 777 modules languages cache > /dev/null
($CURL -e "http://$DOMAIN/install/upgrade.php?language=local" -F 'submit=yes' "http://$DOMAIN/install/upgrade.php?language=local") 2> /dev/null > /dev/null;
$MV settings.php.bak settings.php > /dev/null
$RM -fr settings.php.default libs/dbconnect.php.default install/ latest.zip > /dev/null
echo "[    OK    ]";

VN:F [1.9.22_1171]
Rating: 5.0/5 (1 vote cast)
VN:F [1.9.22_1171]
Rating: 0 (from 0 votes)

Сравнително добро заместване на "order by rand()" в MySQL

Метод 1

select @rand:=CEIL(MAX(uid)*RAND()) AS uid from tablica;select ddate,title,url,visits from tablica where uid>=@rand order by uid limit 10;

Метод 2

SELECT uid,ddate,title,url,visits  FROM tablica T JOIN (SELECT FLOOR(MAX(uid)*RAND()-1024) AS ID FROM tablica) AS x ON T.uid >= x.ID LIMIT 10;

Метод 3

SELECT uid,ddate,title,url,visits FROM tablica T JOIN (SELECT MAX(uid) AS ID FROM tablica) AS x ON T.uid >= FLOOR(x.ID*RAND()) LIMIT 10;

Метод 4

SELECT uid,ddate,title,url,visits FROM tablica t JOIN (SELECT(FLOOR(max(uid) * rand())) as maxid FROM tablica) as tt on t.uid >= tt.maxid LIMIT 10;

VN:F [1.9.22_1171]
Rating: 5.0/5 (1 vote cast)
VN:F [1.9.22_1171]
Rating: 0 (from 0 votes)

Машини за съхранение в MySQL

MySQL предоставя избор между няколко машини за съхранение (storage engines) при създаване на таблица. Различните типове таблици се управляват от различни машини за съхранение, всяка от които има собствени, специфични характеристики.

Ето няколко машини за съхранение:

Машина за съхранение MyISAM
Машина за съхранение MERGE
Машина за съхранение InnoDB
Машина за съхранение MEMORY
Машина за съхранение FEDERATED

Всички таблици, които се управляват от съръва на MySQL , имат редица общи особености. Например всяка таблица в базата данни притежава файл във формата (.frm) в директорията за бази данни. Той съхранява дефиницията за структурата на таблицата и се създава от сървъра. Таблиците имат помежду си и редица различия, които са тясно свързани с машините за съхранение, които сървърът използва за управление на съдържанието им. Всяка от тях има собствен набор от функционални характеристики. Например някои машини създават допълнителни файлове на твърдия диск за управление на данни и за съхранение на индекси, които придружават тези с разширение frm, но типовете им се различават в зависимост от съответната машина за съхранение. Последните се различават и по други особености, например по начина, по който използват заключване за контролиране на съревнованието между заявките, както и по вида на таблиците, които осигуряват: те могат да бъдат транзакционни и нетранзакционни. Споменатите свойства на машините за съхранение имат съществено значение за производителността при обработката на заявки, за контролиране на съревнованието между тях, както и за предотвратяване на т.нар. “мъртва хватка” (Тя се получава в резултат от блокиране на голям брой заявки, които не могат да продължат изпълнението си, тъй като всяка от тях изчаква приключването на останалите).

При създаване на таблица можете да изберете коя машина за съхранение да използвате. Обикновено при избора се гледа коя от тях предлага възможности, които най-добре отговарят на нуждите на конкретното приложение. Преди да направите избора си, трябва да отговорите на въпроса за какъв тип заявки възнамерявате да използвате съответната таблица. След това изберете машина за съхранение, която използва подходящо ниво на заключване. Заключването на ниво таблица, което се използва от MyISAM, работи най-ефективно при комбинации от заявки, осъществяващи интензивно извличане на данни и малко на брой актуализации. Ако желаете да обработвате комбинации от заявки, включващи множество актуализации, е далеч по-добре да използвате innoDB. За нея е характерно заключване на ниво ред и поддържане на множество версии, което осигурява отлична съгласуваност при комбинация от извличане и актуализации на данни. В този случай една заявка може да актуализира дадени редове от таблицата, докато други четат или актуализират останалите.
За да зададете изрично машина за съхранение, можете да използвате опцията engine в конструкцията create table. Дадената по-долу служи за създаване на таблица t с машина за съхранение InnoDB:

CREATE TABLE t (i int) ENGINE = InnoDB;

При липсата на engine, сървърът на MySQL ще създаде таблицата с подразбиращата се машина за съхранение. Тя се определя от стойността на системната променлива storage engine.
За да определите коя машина е използвана за дадена таблица, можете да използвате конструкцията SHOW create table или SHOW table status:
mysql> SHOW CREATE TABLE City\G
*************************** 1. row ***************************
Table: City
Create Table: CREATE TABLE 'City" (
'id' int(ll) not NULL auto_increment,
'Name' char(35) not null default
'CountryCode' char(3) NOT null default
'District' char (20) not null default V,
'Population' int(ll) not null default '0',
PRIMARY key ('id') ) ENGINE=MyISAM DEFAULT CHARSET=latinl
1 row in set (0.00 sec)

mysql> SHOW TABLE STATUS LIKE ' CountryLanguage'\G
*************************** 1 row ***************************
Name: CountryLanguage
Engine: MyiSAM
Version: 10
Row_format: Fixed
Rows: 984
A.vg row length: 39
Data length: 38376
ix data length: 167503724543
Index length: 22528
Data free: 0
auto increment: NULL
Create time: 2005-04-26 22: 15: 35
Update time: 2005-04-26 22: 15: 43
Check time: NULL
Collation: latinl Swedish ci
Checksum: NULL
Create options:
Comment:
row in set (0. 00 sec)

Таблицата informationschema tables съдържа информация също и за машините за съхранение:
mysql> SELECT TABLE_NAME, ENGINE FROM INFORMATION_SCHEMA.TABLES
-> WHERE TABLE SCHEMA = "world';

-----------------------------------------------
TABLE_NAME 1 ENGINE

City 1 MyISAM
Country 1 MylSAM
CountryLanguage MylSAM
------------------------------------------------

Въпреки че имате избор коя машина за съхранение да използвате за дадена таблица, в повечето случаи начинът на използването й след нейното създаване не зависи от машината. Действията върху всички видове таблици се изпълняват с помощта на интерфейса на SQL, a MySQL автоматично управлява детайлите, зависими от машината за съхранение, на по-ниско ниво в архитектурата си. С други думи, вашият интерфейс за контролиране на таблиците се намира в по-високия слой SQL, докато детайлите за управлението им са на по-ниския слой на машината за съхранение. Независимо от това има случаи, в които да знаете коя машина управлява дадената таблица ще ви позволи да я използвате по-ефективно. Оптимизациите, свързани с машините за съхранение, са разгледани в глава 38, „Оптимизиране на бази данни".
За да може да се използва дадена машина за съхранение, тя трябва да се компилира в сървъра и да бъде разрешена. Той използва модулна архитектура: всяка машина за съхранение представлява софтуерен модул, който се компилира в сървъра. Използването на този подход позволява включване само на някои машини в сървъра по време на конфигурирането.
Някои машини за съхранение са винаги налични. Такива са например MylSAM, merge и memory. Други са незадължителни. Поддръжката за тях може да се избере по време на конфигурирането и изграждането на MySQL. Вътрешно компилираните незадължителни машини за съхранение могат да се активират и дезактивират също и чрез стартова опция на сървъра. Например InnoDB е включена във всички двоични пакети. Ако изградите MySQL от сорс код, InnoDB се включва по подразбиране, освен ако не зададете конфигуриращата опция —without-innodb. На всеки сървър, поддръжката на включена машина за съхранение може да се дезактивира при стартиране, с помощта на опцията

—skip-innodb.

Ако желаете да намалите размера на използваната памет, избягвайте конфигурирането на ненужни машини за съхранение в сървъра. За тази цел трябва да компилирате MySQL от сорс код, вместо да използвате предварително подготвен двоичен пакет. В случай, че използвате такъв, който включва вътрешно компилирани незадължителни машини за съхранение, които не са ви необходими, можете да ги дезактивирате по време на работа.
За да видите кои машини са компилирани в сървъра и дали са достъпни по време на работа, можете да използвате конструкцията show engines.

mysql> SHOW ENGINES\G
*************************** 1 row *************************** Engine: MyiSAM Support: DEFAULT
Comment: Default engine as of MySQL 3.23 with great performance
*************************** 2. row ***************************
Engine: MEMORY Support: YES
Comment: Hash based, stored in memory, useful for temporary tables
*************************** 3. row ***************************
Engine: HEAP Support: YES
Comment: Alias for MEMORY

Източник: MySQL 5.0(Пол Дюбоа, Стефан Хинц, Карстен Педерсън)

VN:F [1.9.22_1171]
Rating: 5.0/5 (1 vote cast)
VN:F [1.9.22_1171]
Rating: 0 (from 0 votes)

Поправяне на проблем с php кодирано- IonCube

На един приятел сайта на сайта му бликна следната грешка:

Site error: the file /var/www/index.php requires the ionCube PHP Loader ioncube_loader_lin_5.2.so to be installed by the site administrator.

Очевидно сайта му е кодиран на IonCube и трябва да се сложи декодера.

Решение:

Трябва да се копират файловете от IonCube ioncube-encoded-file.php ioncube-loader-helper.php в уеб директорията където са кодираните php-ta и всичко ще си е наред.

VN:F [1.9.22_1171]
Rating: 5.0/5 (1 vote cast)
VN:F [1.9.22_1171]
Rating: 0 (from 0 votes)

Samba конфигурационен файл за простички цели

Идеята тук е тази самба да споделя една папка по мрежата където всеки да може да пише вътре(най- елементарен конф). Ето малко повече инфо за тези които не са запознати със Samba т.е. ще опиша най- популярните неща за които се ползва Samba:

1. Споделя не ресурс
1.1 Споделяне на папка за запис
1.2 Споделя не папка само за четене
1.3 Споделяне на принтер

Ето примерен Samba (smb.conf) конфигурационен файл за споделяне на папка за запис:

[global]
# Change this to the workgroup/NT-domain name your Samba server will part of
workgroup = dhnet

# server string is the equivalent of the NT Description field
server string = %h server (Samba %v)

# that connects
log file = /var/log/samba/log.%m

# Put a capping on the size of the log files (in Kb).
max log size = 1000
syslog = 0
security = share
# Do something sensible when Samba crashes: mail the admin a backtrace
panic action = /usr/share/samba/panic-action %d
# 'encrypt passwords' in the smb.conf(5) manpage before enabling.
encrypt passwords = true
socket options = TCP_NODELAY

[write]
comment = write
browseable = no
path = /samba/write/
public = yes
writable = yes
guest ok = yes
# Windows clients look for this share name as a source of downloadable

VN:F [1.9.22_1171]
Rating: 5.0/5 (1 vote cast)
VN:F [1.9.22_1171]
Rating: 0 (from 0 votes)

Конфигуриране на DHCP под GNU/Linux Debian

Тук идея е простичка. Един DHCP демон да раздава IP адреси на всеки компютър в конкретната физическа мрежа и възможност конкретни компютри(различават се по MAC-овете си) да получават винаги един и същи IP адрес.

Ето го и DHCP конфигурационния файл:

### dhcp conf written by amri 2005

max-lease-time 7200;

default-lease-time 3600;

option domain-name-servers 8.8.8.8,8.8.4.4;

option domain-name "domain.com";

option time-offset 2;

ddns-update-style none;

####

shared-network tk

{

#########  NETWORKS  #####

subnet 10.10.10.0 netmask 255.255.255.0{

range 10.10.10.2 10.10.10.50;

option subnet-mask 255.255.255.0;

option broadcast-address 10.10.10.255;

option routers 10.10.10.1;

}

########## CLIENTS #######

host PC_stacionaren {fixed-address 10.10.10.4; hardware ethernet 00:0E:2E:3D:B3:C3;}

host PC_notebook {fixed-address 10.10.10.5; hardware ethernet 00:0A:48:1A:17:B6;}

}

VN:F [1.9.22_1171]
Rating: 5.0/5 (1 vote cast)
VN:F [1.9.22_1171]
Rating: 0 (from 0 votes)

Shaper под Debian или ограничаване на Интернет/Скорост по IP

Ето последния cbq.init 7.3

#!/bin/bash
#
#    cbq.init v0.7.3
#    Copyright (C) 1999  Pavel Golubev <[email protected]>
#    Copyright (C) 2001-2004  Lubomir Bulej <[email protected]>
#    Patches for Debian (largely cosmetic) copyright (C) 2000, 2001, 2002
#    Colin Watson <[email protected]>
#    More patches for Debian (largely functional) copyright (C) 2003, 2004
#    David B. Harris <[email protected]>
#
#    chkconfig:   2345 11 89
#    description: sets up CBQ-based traffic control
#
#    This program is free software; you can redistribute it and/or modify
#    it under the terms of the GNU General Public License as published by
#    the Free Software Foundation; either version 2 of the License, or
#    (at your option) any later version.
#
#    This program is distributed in the hope that it will be useful,
#    but WITHOUT ANY WARRANTY; without even the implied warranty of
#    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
#    GNU General Public License for more details.
#
#    You should have received a copy of the GNU General Public License
#    along with this program; if not, write to the Free Software
#    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
#
#    To get the latest version, check on Freshmeat for actual location:
#
#		http://freshmeat.net/projects/cbq.init
#
#
# VERSION HISTORY
# ---------------
# v0.7.3- Deepak Singhal <singhal at users.sourceforge.net>
#	  - fix timecheck to not ignore regular TIME rules after
#	    encountering a TIME rule that spans over midnight
#	- Nathan Shafer <nicodemus at users.sourceforge.net>
#	  - allow symlinks to class files
#	- Seth J. Blank <antifreeze at users.sourceforge.net>
#	  - replace hardcoded ip/tc location with variables
#	- Mark Davis <mark.davis at gmx.de>
#	  - allow setting of PRIO_{MARK,RULE,REALM} in class file
#	- Fernando Sanch <toptnc at users.sourceforge.net>
#	  - allow underscores in interface names
# v0.7.2- Paulo Sedrez
#	  - fix time2abs to allow hours with leading zero in TIME rules
#	- Svetlin Simeonov <zvero at yahoo.com>
#	  - fix cbq_device_list to allow VLAN interfaces
#	- Mark Davis <mark.davis at gmx.de>
#	  - ignore *~ backup files when looking for classes
#	- Mike Boyer <boyer at administrative.com>
#	  - fix to allow arguments to be passed to "restart" command
# v0.7.1- Lubomir Bulej <pallas at kadan.cz>
#	  - default value for PERTURB
#	  - fixed small bug in RULE parser to correctly parse rules with
#	    identical source and destination fields
#	  - faster initial scanning of DEVICE fields
# v0.7	- Lubomir Bulej <pallas at kadan.cz>
#	  - lots of various cleanups and reorganizations; the parsing is now
#	    some 40% faster, but the class ID must be in range 0x0002-0xffff
#	    (again). Because of the number of internal changes and the above
#	    class ID restriction, I bumped the version to 0.7 to indicate
#	    something might have got broken :)
#	  - changed PRIO_{U32,FW,ROUTE} to PRIO_{RULE,MARK,REALM}
#	    for consistency with filter keywords
#	  - exposed "compile" command
#	- Catalin Petrescu <taz at dntis.ro>
#	  - support for port masks in RULE (u32) filter
#	- Jordan Vrtanoski <obeliks at mt.net.mk>
#	  - support for week days in TIME rules
# v0.6.4- Lubomir Bulej <pallas at kadan.cz>
#	  - added PRIO_* variables to allow easy control of filter priorities
#	  - added caching to speed up CBQ start, the cache is invalidated
#	    whenever any of the configuration files changes
#	  - updated the readme section + some cosmetic fixes
# v0.6.3- Lubomir Bulej <pallas at kadan.cz>
#	  - removed setup of (unnecessary) class 1:1 - all classes
#	    now use qdisc's default class 1:0 as their parent
#	  - minor fix in the timecheck branch - classes
#	    without leaf qdisc were not updated
#	  - minor fix to avoid timecheck failure when run
#	    at time with minutes equal to 08 or 09
#	  - respect CBQ_PATH setting in environment
#	  - made PRIO=5 default, rendering it optional in configs
#	  - added support for route filter, see notes about REALM keyword
#	  - added support for fw filter, see notes about MARK keyword
#	  - added filter display to "list" and "stats" commands
#	  - readme section update + various cosmetic fixes
# v0.6.2- Catalin Petrescu <taz at dntis.ro>
#	  - added tunnels interface handling
# v0.6.1- Pavel Golubev <pg at ksi-linux.com>
#	  - added sch_prio module loading
#	    (thanks johan at iglo.virtual.or.id for reminding)
#	  - resolved errors resulting from stricter syntax checking in bash2
#	- Lubomir Bulej <pallas at kadan.cz>
#	  - various cosmetic fixes
# v0.6	- Lubomir Bulej <pallas at kadan.cz>
#	  - attempt to limit number of spawned processes by utilizing
#	    more of sed power (use sed instead of grep+cut)
#	  - simplified TIME parser, using bash builtins
#	  - added initial support for SFQ as leaf qdisc
#	  - reworked the documentation part a little
#	  - incorporated pending patches and ideas submitted by
#	    following people for versions 0.3 into version 0.6
#	- Miguel Freitas <miguel at cetuc.puc-rio.br>
#	  - in case of overlapping TIME parameters, the last match is taken
#	- Juanjo Ciarlante <jjo at mendoza.gov.ar>
#	  - chkconfig tags, list + stats startup parameters
#	  - optional tc & ip command logging (into /var/run/cbq-*)
#	- Rafal Maszkowski <rzm at icm.edu.pl>
#	  - PEAK parameter for setting TBF's burst peak rate
#	  - fix for many config files (use find instead of ls)
# v0.5.1- Lubomir Bulej <pallas at kadan.cz>
#	  - fixed little but serious bug in RULE parser
# v0.5	- Lubomir Bulej <pallas at kadan.cz>
#	  - added options PARENT, LEAF, ISOLATED and BOUNDED. This allows
#	    (with some attention to config file ordering) for creating
#	    hierarchical structures of shapers with classes able (or unable)
#	    to borrow bandwidth from their parents.
#	  - class ID check allows hexadecimal numbers
#	  - rewritten & simplified RULE parser
#	  - cosmetic changes to improve readability
#	  - reorganization to avoid duplicate code (timecheck etc.)
#	  - timecheck doesn't check classes without TIME fields anymore
# v0.4  - Lubomir Bulej <pallas at kadan.cz>
#	  - small bugfix in RULE parsing code
#	  - simplified configuration parsing code
#	  - several small cosmetic changes
#	  - TIME parameter can be now specified more than once allowing you to
#	    differentiate RATE throughout the whole day. Time overlapping is
#	    not checked, first match is taken. Midnight wrap (eg. 20:00-6:00)
#	    is allowed and taken care of.
# v0.3a4- fixed small bug in IF operator. Thanks to
#	  Rafal Maszkowski <rzm at icm.edu.pl>
# v0.3a3- fixed grep bug when using more than 10 eth devices. Thanks to David
#	  Trcka <trcka at poda.cz>.
# v0.3a2- fixed bug in "if" operator. Thanks kad at dgtu.donetsk.ua.
# v0.3a - added TIME parameter. Example: TIME=00:00-19:00;64Kbit/6Kbit
#	  So, between 00:00 and 19:00 the RATE will be 64Kbit.
#	  Just start "cbq.init timecheck" periodically from cron
#	  (every 10 minutes for example). DON'T FORGET though, to run
#	  "cbq.init start" for CBQ to initialize.
# v0.2  - Some cosmetic changes. Now it is more compatible with old bash
#	  version. Thanks to Stanislav V. Voronyi <stas at cnti.uanet.kharkov.ua>.
# v0.1  - First public release
#
#
# README
# ------
#
# First of all - this is just a SIMPLE EXAMPLE of CBQ power.
# Don't ask me "why" and "how" :)
#
# This script is meant to simplify setup and management of relatively simple
# CBQ-based traffic control on Linux. Access to advanced networking features
# of Linux kernel is provided by "ip" and "tc" utilities from A. Kuznetsov's
# iproute2 package, available at ftp://ftp.inr.ac.ru/ip-routing. Because the
# utilities serve primarily to translate user wishes to RTNETLINK commands,
# their interface is rather spartan, intolerant and requires quite a lot of
# typing. And typing is what this script attempts to reduce :)
#
# The advanced networking stuff in Linux is pretty flexible and this script
# aims to bring some of its features to the not-so-hard-core Linux users. Of
# course, there is a tradeoff between simplicity and flexibility and you may
# realize that the flexibility suffered too much for your needs -- time to
# face "ip" and "tc" interface.
#
# To speed up the "start" command, simple caching was introduced in version
# 0.6.4. The caching works so that the sequence of "tc" commands for given
# configuration is stored in a file (/var/cache/cbq.init by default) which
# is used next time the "start" command is run to avoid repeated parsing of
# configuration files. This cache is invalidated whenever any of the CBQ
# configuration files changes. If you want to run "shaper start" without
# caching, run it as "shaper start nocache". If you want to force cache
# invalidation, run it as "shaper start invalidate". Caching is disabled
# if you have logging enabled (ie. CBQ_DEBUG is not empty).
#
# If you only want cqb.init to translate your configuration to "tc" commands,
# use "compile" command which will output "tc" commands required to build
# your configuration. Bear in mind that "compile" does not check if the "tc"
# commands were successful - this is done (in certain places) only when the
# "start nocache" command is used, which is also useful when creating the
# configuration to check whether it is completely valid.
#
# All CBQ parameters are valid for Ethernet interfaces only, The script was
# tested on various Linux kernel versions from series 2.1 to 2.4 and several
# distributions with KSI Linux (Nostromo version) as the premier one.
#
#
# HOW DOES IT WORK?
# -----------------
#
# Every traffic class must be described by a file in the $CBQ_PATH directory
# (/etc/shaper by default) - one file per class.
#
# The config file names must obey mandatory format: cbq-<clsid>.<name> where
# <clsid> is two-byte hexadecimal number in range <0002-FFFF> (which in fact
# is a CBQ class ID) and <name> is the name of the class -- anything to help
# you distinguish the configuration files. For small amount of classes it is
# often possible (and convenient) to let <clsid> resemble bandwidth of the
# class.
#
# Example of valid config name:
#	cbq-1280.My_first_shaper
#
#
# The configuration file may contain the following parameters:
#
### Device parameters
#
# DEVICE=<ifname>,<bandwidth>[,<weight>]	mandatory
# DEVICE=eth0,10Mbit,1Mbit
#
#	<ifname> is the name of the interface you want to control
#		traffic on, e.g. eth0
#	<bandwidth> is the physical bandwidth of the device, e.g. for
#		ethernet 10Mbit or 100Mbit, for arcnet 2Mbit
#	<weight> is tuning parameter that should be proportional to
#		<bandwidth>. As a rule of thumb: <weight> = <bandwidth> / 10
#
# When you have more classes on one interface, it is enough to specify
# <bandwidth> [and <weight>] only once, therefore in other files you only
# need to set DEVICE=<ifname>.
#
### Class parameters
#
# RATE=<speed>					mandatory
# RATE=5Mbit
#
#	Bandwidth allocated to the class. Traffic going through the class is
#	shaped to conform to specified rate. You can use Kbit, Mbit or bps,
#	Kbps and Mbps as suffices. If you don't specify any unit, bits/sec
#	are used. Also note that "bps" means "bytes per second", not bits.
#
# WEIGHT=<speed> 				mandatory
# WEIGHT=500Kbit
#
#	Tuning parameter that should be proportional to RATE. As a rule
#	of thumb, use WEIGHT ~= RATE / 10.
#
# PRIO=<1-8>					optional, default 5
# PRIO=5
#
#	Priority of class traffic. The higher the number, the lesser
#	the priority. Priority of 5 is just fine.
#
# PARENT=<clsid>				optional, default not set
# PARENT=1280
#
#	Specifies ID of the parent class to which you want this class be
#	attached. You might want to use LEAF=none for the parent class as
#	mentioned below. By using this parameter and carefully ordering the
#	configuration files, it is possible to create simple hierarchical
#	structures of CBQ classes. The ordering is important so that parent
#	classes are constructed prior to their children.
#
# LEAF=none|tbf|sfq				optional, default "tbf"
#
#	Tells the script to attach specified leaf queueing discipline to CBQ
#	class. By default, TBF is used. Note that attaching TBF to CBQ class
#	shapes the traffic to conform to TBF parameters and prevents the class
#	from borrowing bandwidth from its parent even if you have BOUNDED set
#	to "no". To allow the class to borrow bandwith (provided it is not
#	bounded), you must set LEAF to "none" or "sfq".
#
#	If you want to ensure (approximately) fair sharing of bandwidth among
#	several hosts in the same class, you might want to specify LEAF=sfq to
#	attach SFQ as leaf queueing discipline to that class.
#
# BOUNDED=yes|no				optional, default "yes"
#
#	If set to "yes", the class is not allowed to borrow bandwidth from
#	its parent class in overlimit situation. If set to "no", the class
#	will be allowed to borrow bandwidth from its parent.
#
# Note:	Don't forget to set LEAF to "none" or "sfq", otherwise the class will
#	have TBF attached to itself and will not be able to borrow unused
#	bandwith from its parent.
#
# ISOLATED=yes|no				optional, default "no"
#
#	If set to "yes", the class will not lend unused bandwidth to
#	its children.
#
### TBF qdisc parameters
#
# BUFFER=<bytes>[/<bytes>]			optional, default "10Kb/8"
#
#	This parameter controls the depth of the token bucket. In other
#	words it represents the maximal burst size the class can send.
#	The optional part of parameter is used to determine the length
#	of intervals in packet sizes, for which the transmission times
#	are kept.
#
# LIMIT=<bytes>					optional, default "15Kb"
#
#	This parameter determines the maximal length of backlog. If
#	the queue contains more data than specified by LIMIT, the
#	newly arriving packets are dropped. The length of backlog
#	determines queue latency in case of congestion.
#
# PEAK=<speed>					optional, default not set
#
#	Maximal peak rate for short-term burst traffic. This allows you
#	to control the absolute peak rate the class can send at, because
#	single TBF that allows 256Kbit/s would of course allow rate of
#	512Kbit for half a second or 1Mbit for a quarter of second.
#
# MTU=<bytes>  					optional, default "1500"
#
#	Maximum number of bytes that can be sent at once over the
#	physical medium. This parameter is required when you specify
#	PEAK parameter. It defaults to MTU of ethernet - for other
#	media types you might want to change it.
#
# Note: Setting TBF as leaf qdisc will effectively prevent the class from
#	borrowing bandwidth from the ancestor class, because even if the
#	class allows more traffic to pass through, it is then shaped to
#	conform to TBF.
#
### SFQ qdisc parameters
#
# The SFQ queueing discipline is a cheap way for sharing class bandwidth
# among several hosts. As it is stochastic, the fairness is approximate but
# it will do the job in most cases. If you want real fairness, you should
# probably use WRR (weighted round robin) or WFQ queueing disciplines. Note
# that SFQ does not do any traffic shaping - the shaping is done by the CBQ
# class the SFQ is attached to.
#
# QUANTUM=<bytes>				optional, default not set
#
#	This parameter should not be set lower than link MTU, for ethernet
#	it is 1500b, or (with MAC header) 1514b which is the value used
#	in Alexey Kuznetsov's examples.
#
# PERTURB=<seconds>				optional, default "10"
#
#	Period of hash function perturbation. If unset, hash reconfiguration
#	will never take place which is what you probably don't want. The
#	default value of 10 seconds is probably a good one.
#
### Filter parameters
#
# RULE=[[saddr[/prefix]][:port[/mask]],][daddr[/prefix]][:port[/mask]]
#
#	These parameters make up "u32" filter rules that select traffic for
#	each of the classes. You can use multiple RULE fields per config.
#
#	The optional port mask should only be used by advanced users who
#	understand how the u32 filter works.
#
# Some examples:
#
#	RULE=10.1.1.0/24:80
#		selects traffic going to port 80 in network 10.1.1.0
#
#	RULE=10.2.2.5
#		selects traffic going to any port on single host 10.2.2.5
#
#	RULE=10.2.2.5:20/0xfffe
#		selects traffic going to ports 20 and 21 on host 10.2.2.5
#
#	RULE=:25,10.2.2.128/26:5000
#		selects traffic going from anywhere on port 50 to
#		port 5000 in network 10.2.2.128
#
#	RULE=10.5.5.5:80,
#		selects traffic going from port 80 of single host 10.5.5.5
#
#
#
# REALM=[srealm,][drealm]
#
#	These parameters make up "route" filter rules that classify traffic
#	according to packet source/destination realms. For information about
#	realms, see Alexey Kuznetsov's IP Command Reference. This script
#	does not define any realms, it justs builds "tc filter" commands
#	for you if you need to classify traffic this way.
#
#	Realm is either a decimal number or a string referencing entry in
#	/etc/iproute2/rt_realms (usually).
#
# Some examples:
#
#	REALM=russia,internet
#		selects traffic going from realm "russia" to realm "internet"
#
#	REALM=freenet,
#		selects traffic going from realm "freenet"
#
#	REALM=10
#		selects traffic going to realm 10
#
#
#
# MARK=<mark>
#
#	These parameters make up "fw" filter rules that select traffic for
#	each of the classes accoring to firewall "mark". Mark is a decimal
#	number packets are tagged with if firewall rules say so. You can
#	use multiple MARK fields per config.
#
#
# Note: Rules for different filter types can be combined. Attention must be
#	paid to the priority of filter rules, which can be set below using
#	PRIO_{RULE,MARK,REALM} variables.
#
### Time ranging parameters
#
# TIME=[<dow>,<dow>, ...,<dow>/]<from>-<till>;<rate>/<weight>[/<peak>]
# TIME=0,1,2,5/18:00-06:00;256Kbit/25Kbit
# TIME=60123/18:00-06:00;256Kbit/25Kbit
# TIME=18:00-06:00;256Kbit/25Kbit
#
#	This parameter allows you to differentiate the class bandwidth
#	throughout the day. You can specify multiple TIME parameters, if
#	the times overlap, last match is taken. The fields <rate>, <weight>
#	and <peak> correspond to parameters RATE, WEIGHT and PEAK (which
#	is optional and applies to TBF leaf qdisc only).
#
#	You can also specify days of week when the TIME rule applies. <dow>
#	is numeric, 0 corresponds to sunday, 1 corresponds to monday, etc.
#
###
#
# Sample configuration file: cbq-1280.My_first_shaper
#
# --------------------------------------------------------------------------
# DEVICE=eth0,10Mbit,1Mbit
# RATE=128Kbit
# WEIGHT=10Kbit
# PRIO=5
# RULE=192.128.1.0/24
# --------------------------------------------------------------------------
#
# The configuration says that we will control traffic on 10Mbit ethernet
# device eth0 and the traffic going to network 192.168.1.0 will be
# processed with priority 5 and shaped to rate of 128Kbit.
#
# Note that you can control outgoing traffic only. If you want to control
# traffic in both directions, you must set up CBQ for both interfaces.
#
# Consider the following example:
#
#                    +---------+      192.168.1.1
# BACKBONE -----eth0-|  linux  |-eth1------*-[client]
#                    +---------+
#
# Imagine you want to shape traffic from backbone to the client to 28Kbit
# and traffic in the opposite direction to 128Kbit. You need to setup CBQ
# on both eth0 and eth1 interfaces, thus you need two config files:
#
# cbq-028.backbone-client
# --------------------------------------------------------------------------
# DEVICE=eth1,10Mbit,1Mbit
# RATE=28Kbit
# WEIGHT=2Kbit
# PRIO=5
# RULE=192.168.1.1
# --------------------------------------------------------------------------
#
# cbq-128.client-backbone
# --------------------------------------------------------------------------
# DEVICE=eth0,10Mbit,1Mbit
# RATE=128Kbit
# WEIGHT=10Kbit
# PRIO=5
# RULE=192.168.1.1,
# --------------------------------------------------------------------------
#
# Pay attention to comma "," in the RULE field - it denotes source address!
#
# Enjoy.
#
#############################################################################

### BEGIN INIT INFO
# Provides:          shaper
# Required-Start:    $local_fs $remote_fs $syslog $network
# Required-Stop:     $local_fs $remote_fs $syslog $network
# Default-Start:     2 3 4 5
# Default-Stop:      0 1 6
# Short-Description: Traffic shaper
### END INIT INFO

export LC_ALL=C

### Command locations
TC=/sbin/tc
IP=/sbin/ip
MP=/sbin/modprobe

DESC="CBQ traffic shaping"

### Default filter priorities (must be different)
PRIO_RULE_DEFAULT=${PRIO_RULE:-100}
PRIO_MARK_DEFAULT=${PRIO_MARK:-200}
PRIO_REALM_DEFAULT=${PRIO_REALM:-300}

### Default CBQ_PATH & CBQ_CACHE settings
CBQ_PATH=${CBQ_PATH:-/etc/shaper}
CBQ_CACHE=${CBQ_CACHE:-/var/cache/shaper/cbq.init}

### Uncomment to enable logfile for debugging
#CBQ_DEBUG="/var/run/cbq-$1"

### Modules to probe for. Uncomment the last CBQ_PROBE
### line if you have QoS support compiled into kernel
CBQ_PROBE="sch_cbq sch_tbf sch_sfq sch_prio"
CBQ_PROBE="$CBQ_PROBE cls_fw cls_u32 cls_route"
#CBQ_PROBE=""

### Keywords required for qdisc & class configuration
CBQ_WORDS="DEVICE|RATE|WEIGHT|PRIO|PARENT|LEAF|BOUNDED|ISOLATED"
CBQ_WORDS="$CBQ_WORDS|PRIO_MARK|PRIO_RULE|PRIO_REALM|BUFFER"
CBQ_WORDS="$CBQ_WORDS|LIMIT|PEAK|MTU|QUANTUM|PERTURB"

. /etc/default/shaper

#############################################################################
############################# SUPPORT FUNCTIONS #############################
#############################################################################

### Get list of network devices
cbq_device_list () {
	ip link show| sed -n "/^[0-9]/ \
		{ s/^[0-9]\+: \([a-z0-9._]\+\)[:@].*/\1/; p; }"
} # cbq_device_list

### Remove root class from device $1
cbq_device_off () {
	tc qdisc del dev $1 root 2> /dev/null
} # cbq_device_off

### Remove CBQ from all devices
cbq_off () {
	for dev in `cbq_device_list`; do
		cbq_device_off $dev
	done
} # cbq_off

### Prefixed message
cbq_message () {
	echo -e "**CBQ: $@" >&2
} # cbq_message

### Failure message
cbq_failure () {
	cbq_message "$@"
	exit 1
} # cbq_failure

### Failure w/ cbq-off
cbq_fail_off () {
	cbq_message "$@"
	cbq_off
	exit 1
} # cbq_fail_off

### Failure message, exiting 0
cbq_soft_failure () {
	quiet=$1
	shift
	[ -n "$quiet" ] || cbq_message "$@"
	exit 0
} # cbq_soft_failure

### Failure w/ cbq-off, exiting 0
cbq_soft_fail_off () {
	quiet=$1
	shift
	[ -n "$quiet" ] || cbq_message "$@"
	cbq_off
	exit 0
} # cbq_soft_fail_off

### Convert time to absolute value
cbq_time2abs () {
	local min=${1##*:}; min=${min##0}
	local hrs=${1%%:*}; hrs=${hrs##0}
	echo $[hrs*60 + min]
} # cbq_time2abs

### Display CBQ setup
cbq_show () {
	for dev in `cbq_device_list`; do
		[ `tc qdisc show dev $dev| wc -l` -eq 0 ] && continue
		echo -e "### $dev: queueing disciplines\n"
		tc $1 qdisc show dev $dev; echo

		[ `tc class show dev $dev| wc -l` -eq 0 ] && continue
		echo -e "### $dev: traffic classes\n"
		tc $1 class show dev $dev; echo

		[ `tc filter show dev $dev| wc -l` -eq 0 ] && continue
		echo -e "### $dev: filtering rules\n"
		tc $1 filter show dev $dev; echo
	done
} # cbq_show

### Check configuration and load DEVICES, DEVFIELDS and CLASSLIST from $1
cbq_init () {
	### Get a list of configured classes
	CLASSLIST=`find $1 \( -type f -or -type l \) -name 'cbq-*' \
		-not -name '*~' -maxdepth 1 -printf "%f\n"| sort`
	[ -z "$CLASSLIST" ] &&
		cbq_soft_failure "$2" "no configuration files found in $1!"

	### Gather all DEVICE fields from $1/cbq-*
	DEVFIELDS=`find $1 \( -type f -or -type l \) -name 'cbq-*' \
		  -not -name '*~' -maxdepth 1| xargs sed -n 's/#.*//; \
		  s/[[:space:]]//g; /^DEVICE=[^,]*,[^,]*\(,[^,]*\)\?/ \
		  { s/.*=//; p; }'| sort -u`
	[ -z "$DEVFIELDS" ] &&
		cbq_failure "no DEVICE field found in $1/cbq-*!"

	### Check for different DEVICE fields for the same device
	DEVICES=`echo "$DEVFIELDS"| sed 's/,.*//'| sort -u`
	[ `echo "$DEVICES"| wc -l` -ne `echo "$DEVFIELDS"| wc -l` ] &&
		cbq_failure "different DEVICE fields for single device!\n$DEVFIELDS"
} # cbq_init

### Load class configuration from $1/$2
cbq_load_class () {
	CLASS=`echo $2| sed 's/^cbq-0*//; s/^\([0-9a-fA-F]\+\).*/\1/'`
	CFILE=`sed -n 's/#.*//; s/[[:space:]]//g; /^[[:alnum:]_]\+=[[:alnum:].,:;/*@-_]\+$/ p' $1/$2`

	### Check class number
	IDVAL=`/usr/bin/printf "%d" 0x$CLASS 2> /dev/null`
	[ $? -ne 0 -o $IDVAL -lt 2 -o $IDVAL -gt 65535 ] &&
		cbq_fail_off "class ID of $2 must be in range <0002-FFFF>!"

	### Set defaults & load class
	RATE=""; WEIGHT=""; PARENT=""; PRIO=5
	LEAF=tbf; BOUNDED=yes; ISOLATED=no
	BUFFER=10Kb/8; LIMIT=15Kb; MTU=1500
	PEAK=""; PERTURB=10; QUANTUM=""

	PRIO_RULE=$PRIO_RULE_DEFAULT
	PRIO_MARK=$PRIO_MARK_DEFAULT
	PRIO_REALM=$PRIO_REALM_DEFAULT

	eval `echo "$CFILE"| grep -E "^($CBQ_WORDS)="`

	### Require RATE/WEIGHT
	[ -z "$RATE" -o -z "$WEIGHT" ] &&
		cbq_fail_off "missing RATE or WEIGHT in $2!"

	### Class device
	DEVICE=${DEVICE%%,*}
	[ -z "$DEVICE" ] && cbq_fail_off "missing DEVICE field in $2!"

	BANDWIDTH=`echo "$DEVFIELDS"| sed -n "/^$DEVICE,/ \
		  { s/[^,]*,\([^,]*\).*/\1/; p; q; }"`

	### Convert to "tc" options
	PEAK=${PEAK:+peakrate $PEAK}
	PERTURB=${PERTURB:+perturb $PERTURB}
	QUANTUM=${QUANTUM:+quantum $QUANTUM}

	[ "$BOUNDED" = "no" ] && BOUNDED="" || BOUNDED="bounded"
	[ "$ISOLATED" = "yes" ] && ISOLATED="isolated" || ISOLATED=""
} # cbq_load_class

#############################################################################
#################################### INIT ###################################
#############################################################################

### Check for presence of ip-route2 in usual place
# Debian change: we don't - this initscript will remain in-place even
# after the package has been removed. So even if its dependencies
# aren't around (iproute namely), we don't 'exit 1', we exit gracefully.
[ -x "$TC" ] && [ -x "$IP" ] || exit 0
# cbq_failure "ip-route2 utilities not installed or executable!"

### ip/tc wrappers
if [ "$1" = "compile" ]; then
	### no module probing
	CBQ_PROBE=""

	ip () {
		$IP "$@"
	} # ip

	### echo-only version of "tc" command
	tc () {
		echo "$TC $@"
	} # tc

elif [ -n "$CBQ_DEBUG" ]; then
	echo -e "# `date`" > $CBQ_DEBUG

	### Logging version of "ip" command
	ip () {
		echo -e "\n# ip $@" >> $CBQ_DEBUG
		$IP "$@" 2>&1 | tee -a $CBQ_DEBUG
	} # ip

	### Logging version of "tc" command
	tc () {
		echo -e "\n# tc $@" >> $CBQ_DEBUG
		$TC "$@" 2>&1 | tee -a $CBQ_DEBUG
	} # tc
else
	### Default wrappers

	ip () {
		$IP "$@"
	} # ip

	tc () {
		$TC "$@"
	} # tc
fi # ip/tc wrappers

case "$1" in

#############################################################################
############################### START/COMPILE ###############################
#############################################################################

start|compile)

[ "$1" = "start" ] && echo -n "Starting $DESC: "

conffiles="$(ls /etc/shaper)"
if [ -z "$conffiles" ]; then
    echo "no configuration files found in /etc/shaper/."
    exit 0
fi

### Probe QoS modules (start only)
### Ignore errors from this, because the user might have these compiled
### into the kernel. We'll see problems later if they're really missing.
if [ -x "$MP" ] && [ -e /proc/modules ]; then
	for module in $CBQ_PROBE; do
		#$MP $module || cbq_failure "failed to load module $module"
		"$MP" $module 2>&1 | grep -v "Can't locate module" >&2
	done
fi

### If we are in compile/nocache/logging mode, don't bother with cache
if [ "$1" != "compile" -a "$2" != "nocache" -a -z "$CBQ_DEBUG" ]; then
	VALID=1

	### validate the cache
	[ "$2" = "invalidate" -o ! -f $CBQ_CACHE ] && VALID=0
	if [ $VALID -eq 1 ]; then
		[ `find $CBQ_PATH -maxdepth 1 -newer $CBQ_CACHE| \
		  wc -l` -gt 0 ] && VALID=0
	fi

	### compile the config if the cache is invalid
	if [ $VALID -ne 1 ]; then
		$0 compile > $CBQ_CACHE ||
			cbq_fail_off "failed to compile CBQ configuration!"
	fi

	### run the cached commands
	exec /bin/sh $CBQ_CACHE
fi

### Load DEVICES, DEVFIELDS and CLASSLIST
cbq_init $CBQ_PATH

### Setup root qdisc on all configured devices
for dev in $DEVICES; do
	### Retrieve device bandwidth and, optionally, weight
	DEVTEMP=`echo "$DEVFIELDS"| sed -n "/^$dev,/ { s/$dev,//; p; q; }"`
	DEVBWDT=${DEVTEMP%%,*};	DEVWGHT=${DEVTEMP##*,}
	[ "$DEVBWDT" = "$DEVWGHT" ] && DEVWGHT=""

	### Device bandwidth is required
	if [ -z "$DEVBWDT" ]; then
		cbq_message "could not determine bandwidth for device $dev!"
		cbq_failure "please set up the DEVICE fields properly!"
	fi

	### Check if the device is there
	ip link show $dev &> /dev/null ||
		cbq_fail_off "device $dev not found!"

	### Remove old root qdisc from device
	cbq_device_off $dev

	### Setup root qdisc + class for device
	tc qdisc add dev $dev root handle 1 cbq \
	bandwidth $DEVBWDT avpkt 1000 cell 8

	### Set weight of the root class if set
	[ -n "$DEVWGHT" ] &&
		tc class change dev $dev root cbq weight $DEVWGHT allot 1514

	[ "$1" = "compile" ] && echo
done # dev

### Setup traffic classes
for classfile in $CLASSLIST; do
	cbq_load_class $CBQ_PATH $classfile

	### Create the class
	tc class add dev $DEVICE parent 1:$PARENT classid 1:$CLASS cbq \
	bandwidth $BANDWIDTH rate $RATE weight $WEIGHT prio $PRIO \
	allot 1514 cell 8 maxburst 20 avpkt 1000 $BOUNDED $ISOLATED ||
		cbq_fail_off "failed to add class $CLASS with parent $PARENT on $DEVICE!"

	### Create leaf qdisc if set
	if [ "$LEAF" = "tbf" ]; then
		tc qdisc add dev $DEVICE parent 1:$CLASS handle $CLASS tbf \
		rate $RATE buffer $BUFFER limit $LIMIT mtu $MTU $PEAK
	elif [ "$LEAF" = "sfq" ]; then
		tc qdisc add dev $DEVICE parent 1:$CLASS handle $CLASS sfq \
		$PERTURB $QUANTUM
	fi

	### Create fw filter for MARK fields
	for mark in `echo "$CFILE"| sed -n '/^MARK/ { s/.*=//; p; }'`; do
		### Attach fw filter to root class
		tc filter add dev $DEVICE parent 1:0 protocol ip \
		prio $PRIO_MARK handle $mark fw classid 1:$CLASS
	done ### mark

	### Create route filter for REALM fields
	for realm in `echo "$CFILE"| sed -n '/^REALM/ { s/.*=//; p; }'`; do
		### Split realm into source & destination realms
		SREALM=${realm%%,*}; DREALM=${realm##*,}
		[ "$SREALM" = "$DREALM" ] && SREALM=""

		### Convert asterisks to empty strings
		SREALM=${SREALM#\*}; DREALM=${DREALM#\*}

		### Attach route filter to the root class
		tc filter add dev $DEVICE parent 1:0 protocol ip \
		prio $PRIO_REALM route ${SREALM:+from $SREALM} \
		${DREALM:+to $DREALM} classid 1:$CLASS
	done ### realm

	### Create u32 filter for RULE fields
	for rule in `echo "$CFILE"| sed -n '/^RULE/ { s/.*=//; p; }'`; do
		### Split rule into source & destination
		SRC=${rule%%,*}; DST=${rule##*,}
		[ "$SRC" = "$rule" ] && SRC=""

		### Split destination into address, port & mask fields
		DADDR=${DST%%:*}; DTEMP=${DST##*:}
		[ "$DADDR" = "$DST" ] && DTEMP=""

		DPORT=${DTEMP%%/*}; DMASK=${DTEMP##*/}
		[ "$DPORT" = "$DTEMP" ] && DMASK="0xffff"

		### Split up source (if specified)
		SADDR=""; SPORT=""
		if [ -n "$SRC" ]; then
			SADDR=${SRC%%:*}; STEMP=${SRC##*:}
			[ "$SADDR" = "$SRC" ] && STEMP=""

			SPORT=${STEMP%%/*}; SMASK=${STEMP##*/}
			[ "$SPORT" = "$STEMP" ] && SMASK="0xffff"
		fi

		### Convert asterisks to empty strings
		SADDR=${SADDR#\*}; DADDR=${DADDR#\*}

		### Compose u32 filter rules
		u32_s="${SPORT:+match ip sport $SPORT $SMASK}"
		u32_s="${SADDR:+match ip src $SADDR} $u32_s"
		u32_d="${DPORT:+match ip dport $DPORT $DMASK}"
		u32_d="${DADDR:+match ip dst $DADDR} $u32_d"

		### Uncomment the following if you want to see parsed rules
		#echo "$rule: $u32_s $u32_d"

		### Attach u32 filter to the appropriate class
		tc filter add dev $DEVICE parent 1:0 protocol ip \
		prio $PRIO_RULE u32 $u32_s $u32_d classid 1:$CLASS
	done ### rule

	[ "$1" = "compile" ] && echo
done ### classfile

if [ "$1" = "start" ]; then
	echo "shaper."
else # compile
	echo 'echo "shaper."'
fi

;;

#############################################################################
################################# TIME CHECK ################################
#############################################################################

timecheck)

### Get time + weekday
TIME_TMP=`date +%w/%k:%M`
TIME_DOW=${TIME_TMP%%/*}
TIME_NOW=${TIME_TMP##*/}

### Load DEVICES, DEVFIELDS and CLASSLIST
cbq_init $CBQ_PATH quiet

### Run through all classes
for classfile in $CLASSLIST; do
	### Gather all TIME rules from class config
	TIMESET=`sed -n 's/#.*//; s/[[:space:]]//g; /^TIME/ { s/.*=//; p; }' \
		$CBQ_PATH/$classfile`
	[ -z "$TIMESET" ] && continue

	MATCH=0; CHANGE=0
	for timerule in $TIMESET; do
		TIME_ABS=`cbq_time2abs $TIME_NOW`

		### Split TIME rule to pieces
		TIMESPEC=${timerule%%;*}; PARAMS=${timerule##*;}
		WEEKDAYS=${TIMESPEC%%/*}; INTERVAL=${TIMESPEC##*/}
		BEG_TIME=${INTERVAL%%-*}; END_TIME=${INTERVAL##*-}

		### Check the day-of-week (if present)
		[ "$WEEKDAYS" != "$INTERVAL" -a \
		  -n "${WEEKDAYS##*$TIME_DOW*}" ] && continue

		### Compute interval boundaries
		BEG_ABS=`cbq_time2abs $BEG_TIME`
		END_ABS=`cbq_time2abs $END_TIME`

		### Midnight wrap fixup
		if [ $BEG_ABS -gt $END_ABS ]; then
			[ $TIME_ABS -le $END_ABS ] &&
				TIME_ABS=$[TIME_ABS + 24*60]

			END_ABS=$[END_ABS + 24*60]
		fi

		### If the time matches, remember params and set MATCH flag
		if [ $TIME_ABS -ge $BEG_ABS -a $TIME_ABS -lt $END_ABS ]; then
			TMP_RATE=${PARAMS%%/*}; PARAMS=${PARAMS#*/}
			TMP_WGHT=${PARAMS%%/*}; TMP_PEAK=${PARAMS##*/}

			[ "$TMP_PEAK" = "$TMP_WGHT" ] && TMP_PEAK=""
			TMP_PEAK=${TMP_PEAK:+peakrate $TMP_PEAK}

			MATCH=1
		fi
	done ### timerule

	cbq_load_class $CBQ_PATH $classfile

	### Get current RATE of CBQ class
	RATE_NOW=`tc class show dev $DEVICE| sed -n \
		 "/cbq 1:$CLASS / { s/.*rate //; s/ .*//; p; q; }"`
	[ -z "$RATE_NOW" ] && continue

	### Time interval matched
	if [ $MATCH -ne 0 ]; then

		### Check if there is any change in class RATE
		if [ "$RATE_NOW" != "$TMP_RATE" ]; then
			NEW_RATE="$TMP_RATE"
			NEW_WGHT="$TMP_WGHT"
			NEW_PEAK="$TMP_PEAK"
			CHANGE=1
		fi

	### Match not found, reset to default RATE if necessary
	elif [ "$RATE_NOW" != "$RATE" ]; then
		NEW_WGHT="$WEIGHT"
		NEW_RATE="$RATE"
		NEW_PEAK="$PEAK"
		CHANGE=1
	fi

	### If there are no changes, go for next class
	[ $CHANGE -eq 0 ] && continue

	### Replace CBQ class
	tc class replace dev $DEVICE classid 1:$CLASS cbq \
	bandwidth $BANDWIDTH rate $NEW_RATE weight $NEW_WGHT prio $PRIO \
	allot 1514 cell 8 maxburst 20 avpkt 1000 $BOUNDED $ISOLATED

	### Replace leaf qdisc (if any)
	if [ "$LEAF" = "tbf" ]; then
		tc qdisc replace dev $DEVICE handle $CLASS tbf \
		rate $NEW_RATE buffer $BUFFER limit $LIMIT mtu $MTU $NEW_PEAK
	fi

	cbq_message "$TIME_NOW: class $CLASS on $DEVICE changed rate ($RATE_NOW -> $NEW_RATE)"
done ### class file
;;

#############################################################################
################################## THE REST #################################
#############################################################################

stop)
	echo -n "Stopping $DESC: "
	cbq_off
	echo "shaper."
	;;

list)
	cbq_show
	;;

stats)
	cbq_show -s
	;;

restart|force-reload)
	shift
	$0 stop
	$0 start "$@"
	;;

*)
	echo "Usage: `basename $0` {start|compile|stop|restart|force-reload|timecheck|list|stats}"
esac

exit 0

Официална страница на CBQ Init: http://cbqinit.sourceforge.net/

Примерен конф за ограничение на три IP в група:

File: /etc/shaper/cbq-00202.server-users202

DEVICE=eth3,100Mbit,20Mbit
RATE=3730Kbit
WEIGHT=10Kbit
PRIO=5
RULE=172.16.10.122
RULE=172.16.30.82
RULE=172.16.20.12

File: /etc/shaper/cbq-00202.users202-server

DEVICE=eth4,100Mbit,20Mbit
RATE=3730Kbit
WEIGHT=10Kbit
PRIO=5
RULE=172.16.10.122
RULE=172.16.30.82
RULE=172.16.20.12

VN:F [1.9.22_1171]
Rating: 5.0/5 (1 vote cast)
VN:F [1.9.22_1171]
Rating: 0 (from 0 votes)

Просто няколко проблема и решения

Проблем:
nf_conntrack: table full, dropping packet

Решение:
net.netfilter.nf_conntrack_max = 65536

Тестване на прокси през конзола

export http_proxy=http://44.44.44.1:3128

wget --proxy http://mpetrov.net

links -http-proxy http://mpetrov.net

Премахване на ^M от Windows-ки файл, чрез VIM

vim dosfile.txt

%s/\r//g

Ето такъв файл със ^M:

VN:F [1.9.22_1171]
Rating: 5.0/5 (1 vote cast)
VN:F [1.9.22_1171]
Rating: 0 (from 0 votes)