proxy代理   soft软件   IT 业界特快   norton 诺顿病毒库   代理列表   search FTP搜索   whois IP地理位置   blog 追求完美  
money理财   life生活   RSS聚合门户   firefox WEB浏览器   免费域名   typeset 假古文   AntiVirus 反病毒   ipcn 站点导航  

« 7-Zip 4.36 Beta 增加 ISO CRC 支持 | Main | [squid] IPCN proxy/accel 升级至 squid-2.5.STABLE13 »

March 13, 2006

squid-2.5.STABLE13 released

版权声明:可以任意转载,转载时请务必以超链接形式标明文章原始出处和作者信息及本声明。
https://windtear.net/archives/2006/03/13/000936.html
http://windtear.net/archives/2006/03/13/000936.html

squid-2.5.STABLE13 released

http://www.squid-cache.org/Versions/v2/2.5/squid-2.5.STABLE13.tar.gz
http://www.squid-cache.org/Versions/v2/2.5/squid-2.5.STABLE13.tar.bz2
http://www.squid-cache.org/Versions/v2/2.5/squid-2.5.STABLE12-STABLE13.diff.gz

http://www.squid-cache.org/Versions/v2/2.5/squid-2.5.STABLE13-RELEASENOTES.html
Key changes squid-2.5.STABLE12 to 2.5.STABLE13:

    * connstate related memory leak plugged
    * ftp upload issues fixed
    * delay pools working better on -k reconfigure
    * error pages translated to Azerbaijani
    * a number of other minor and cosmetic bugfixes.

http://www.squid-cache.org/Versions/v2/2.5/ChangeLog.txt
Changes to squid-2.5.STABLE12 (12 Mar 2006)
	- [Minor] Fails to compile on Solaris and some other platforms
	  with undefined reference to setenv (Bug #1435)
	- [Cosmetic] Added WebDAV REPORT method to know HTTP methods list
	- [Minor] Squid ntlm_auth (not the Samba provided one) giving
	  odd results if --enable-ntlm-fail-open is used (Bug #1022)
	- [Minor] wbinfo_group.pl doesn't work with Samba 3.0.21 and later
	  (Bug #1472)
	- [Minor] Squid crash when asyncio function counters url accessed
	  from Cachemgr CGI (Bug #1464)
	- [Cosmetic] Linix compile warning about prctl called with too few
 	  arguments (Bug #1483)
	- [Minor] Wrong timezone declaration for 64 bit Irix (Bug #1479)
	- [Minor] Some 206 responses logged incorrectly (Bug #1511)
	- [Minor] Issues in processing ranges on objects >2GB (Bug #437)
	- [Cosmetic] Segmentation fault on empty proxy_auth ACLs (Bug #1414)
	- [Minor] Ident access lists don't work in delay_access statements
	  (Bug #1428)
	- [Minor] Some clients support NTLM even if not initially negotiating
	  persistent connections (Bug #1447)
	- [Medium] 504 Gateway Time-out on FTP uploads (Bug #1459)
	- [Medium] delay pools given too much bandwidht after "-k reconfigure"
	  (Bug #1481)
	- [Cosmetic] New persistent_connection_after_error configuration
	  directive (Bug #1482)
	- [Cosmetic] Hangs at 100% CPU if /dev/null is not accessible (Bug
	  #1484)
	- [Minor] Fails to compile on Fedora Core 5 test 2 x86_64 (Bug #1492)
	- [Cosmetic] Typo in ftp.c (Bug #1507)
	- [Cosmetic] Error in FTP listings of files with -> in their name
	  (Bug #1508)
	- [Cosmetic] With Squid-2.5 there is no more the DUPLICATE IP logging
 	  in cache.log (Bug #779)
	- [Minor] Fails to process long host names (Bug #1434)
	- [Cosmetic] Azerbaijani errors translation (Bug #1454)
	- [Cosmetic] misleading error message message for bad/unresolveable
	  cache_peer name (Bug #1504)
	- [Cosmetic] confusing statistics on stateful helpers (NTLM auth)
	  (Bug #1506)
	- [Major] connstate memory leak (Bug #1522)

http://www.squid-cache.org/Versions/v2/2.5/bugs/#squid-2.5.STABLE12-connstate_leak
http://www.squid-cache.org/Versions/v2/2.5/bugs/squid-2.5.STABLE12-connstate_leak.patch

connstate memory leak
synopsis 	connstate memory leak on cetain failed requests
severity 	Major
date 	2006-03-10 23:17
bugzilla 	#1522
versions 	Squid-2.5 and earlier
platforms 	All
patch 	squid-2.5.STABLE12-connstate_leak.patch

http://www.squid-cache.org/bugs/show_bug.cgi?id=1522
Description:  [reply]  	 Opened: 2006-03-08 02:08

squid 2.5 stable12 is constantly growing in memory size.
cachemgr reports that cbdata connstate is the offender:

cbdata ConnStateData (4)        184     6070727 1090834 1090834 0.00    88     
6070727 1090834 1090834 100

squid is using 1.090.834 KB for connstate data, wich is 88% of its total
memory.
bow has 2GB of ram but is swapping. furthermore, this number is constantly
growing.
and squid has only been up for 2 days:
Start Time:     Mon, 06 Mar 2006 10:14:43 GMT
Current Time:   Wed, 08 Mar 2006 09:04:38 GMT

Nice part is that we have 2 exactly identical boxen and that this problem hits
only one of them. clients are distributed amongs them by round-robin dns I am
sure that one client use is causing this but finding it is rather difficult.

what could I send more to help on this ?
本blog WWW

Posted by windtear at March 13, 2006 11:51 PM

本站使用中的任何问题,请与 windtear @ windtear.net 联系
Copyright© 1999-2024 Windtear. All rights reserved.