[squid-users] Squid Tune on high traffic
Hi, I am using squid on high traffic netwok (50 mbps). Squid working fine for 1 to 2 hour after that al the pages become slow including the cached page. If i do netstat at that time, there was lot of SYN_WAIT and FIN_WAIT1 with lot of send queue. Is there any where i can handle to close this SYN_WAIT and FIN_WAIT1 in 15 to 30 Secs. Because once i stop squid and all the connections are flushed, its started working fine for another 2 hours. Regards Chima
Re: [squid-users] Squid Tune on high traffic
Hi OS - debian 2.2 , Kernel-2.2.x Squid Version - 2.4 Regards Sathyan On 3/12/06, Mark Elsen <[EMAIL PROTECTED]> wrote: > > Hi, > > > > I am using squid on high traffic netwok (50 mbps). Squid working fine for 1 > > to 2 hour after that al the pages become slow including the cached page. > > > > If i do netstat at that time, there was lot of SYN_WAIT and FIN_WAIT1 with > > lot of send queue. > > > > Is there any where i can handle to close this SYN_WAIT and FIN_WAIT1 in 15 > > to 30 Secs. Because once i stop squid and all the connections are flushed, > > its started working fine for another 2 hours. > > > > Regards > > Chima > > > > - Squid version ? > - >OS.platform.version< (!) > > M. >
[squid-users] Transparent proxy with tproxy
Hi, I have configured transparent proxy wit tproxy,iptables. I have 2 problems in this setup First one is, i am getting the following msg in the cache log tproxy ip=192.168.5.1,0x18a0b1ca,port=0 ERROR ASSIGN can anybody encountered this error msg and what is the meaning of this error msg and when it will come, Second issue is after enabling the cache for 2 mins browsing is very solw some times the page is not opening and after 2 mins browsing is normal. Is it anything to do with cache or ip_conntrack? Thanks and Regards Chima
[squid-users] Transparent proxy with tproxy
Hi, I have configured transparent proxy wit tproxy,iptables. I have 2 problems in this setup First one is, i am getting the following msg in the cache log tproxy ip=192.168.5.1,0x18a0b1ca,port=0 ERROR ASSIGN can anybody encountered this error msg and what is the meaning of this error msg and when it will come, Second issue is after enabling the cache for 2 mins browsing is very solw some times the page is not opening and after 2 mins browsing is normal. Is it anything to do with cache or ip_conntrack? Thanks and Regards Chima
Re: [squid-users] Transparent proxy with tproxy
Hi, There was msg in /var/log/messages like Kernel:ip_conntrack table full droping packet Thanks and Regards Chima On 6/1/06, Henrik Nordstrom <[EMAIL PROTECTED]> wrote: tor 2006-06-01 klockan 15:32 +0530 skrev chima s: > Second issue is after enabling the cache for 2 mins browsing is very > solw some times the page is not opening and after 2 mins browsing is > normal. > > Is it anything to do with cache or ip_conntrack? Anything in /var/log/messages? Regards Henrik -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.3 (GNU/Linux) iD8DBQBEf3keB5pTNio2V7IRAg/bAJ97uB7huWNJz4ktpEcrhwgNzU1BUgCcC1li +6TVFUWB/lR9SRtehfjy+bw= =xuzR -END PGP SIGNATURE-
Re: [squid-users] Transparent proxy with tproxy
Hi, What will be rule to redirect the port 80 traffic to sqiud with tproxy table. I am using kernel-2.6.10 and iptables 1.3.3 and patched bith the kernel and iptables utility, but still i am getting error if i apply the rule iptables -t tproxy -A PREROUTING -j TPROXY --on-port 3128 Unknown arg `--on-port' iptables -t nat -A PREROUTING -j DNAT --to-dest 172.16.8.5 --to-port 3128 Unknown arg `--to-port' Regards Chima On 6/2/06, Steven Wilton <[EMAIL PROTECTED]> wrote: The TPROXY_ASSIGN message is indicating that squid has asked the kernel to assign the client's ip address to the server-side connection, and the kernel has returned an error. The kernel needs to be patched with the kernel-side tproxy patch, otherwise this will always fail. Iptables must also be patched to support tproxy, and you must use the TPROXY iptables target, and "-t tproxy" (instead of the REDIRECT with "-t nat"). These patches can be found at http://www.balabit.com/downloads/tproxy/ These errors are not fatal, they just indicate that squid is not spoofing the client's ip address. Regards Steven > -Original Message- > From: chima s [mailto:[EMAIL PROTECTED] > Sent: Thursday, 1 June 2006 6:02 PM > To: squid-users@squid-cache.org > Subject: [squid-users] Transparent proxy with tproxy > > Hi, > > I have configured transparent proxy wit tproxy,iptables. I have 2 > problems in this setup > > First one is, i am getting the following msg in the cache log > > tproxy ip=192.168.5.1,0x18a0b1ca,port=0 ERROR ASSIGN > > can anybody encountered this error msg and what is the meaning of this > error msg and when it will come, > > > Second issue is after enabling the cache for 2 mins browsing is very > solw some times the page is not opening and after 2 mins browsing is > normal. > > Is it anything to do with cache or ip_conntrack? > > Thanks and Regards > Chima > > -- > No virus found in this incoming message. > Checked by AVG Free Edition. > Version: 7.1.394 / Virus Database: 268.8.0/353 - Release > Date: 31/05/2006 > > -- No virus found in this outgoing message. Checked by AVG Free Edition. Version: 7.1.394 / Virus Database: 268.8.1/354 - Release Date: 1/06/2006
[squid-users] cachemgr stat
Hi, I am using the squid as transparent cache with tproxy and configured diskd as a storage system in high network traffic. The Squid using 99% of CPU. How do i find which is taking more CPU through cachemgr statstic. Regards Chima
Re: [squid-users] TPROXY on squid-2.6S1
hi, You have to patch the kernel and iptables before compiling squid with tproxy. You can download the patch from http://www.balabit.com/downloads/tproxy/ Regards CS On 7/5/06, Kashif Ali Bukhari <[EMAIL PROTECTED]> wrote: Salam, i am trying to compile squid on fedoracor 3 with tproxy and get this error while configuration checking if TPROXY header files are installed... no WARNING: Cannot find TPROXY headers, you need to install the tproxy package from: - lynx http://www.balabit.com/downloads/tproxy/ Any help ? -- Syed Kashif Ali Bukhari +92-300-4295604 Network Engineer Beaconhouse IT Services, Lahore Pakistan
[squid-users] coss configuration example
Hi, Can any one give the squid configuration to use coss cache_dir. My system info Fedora Core release 3 Kernel : 2.6.10 Squid : squid-2.6.STABLE2-20060804 MEM: 2GB HD : 2*72 GB SCSI System Filesystem: reiserfs How COSS_MEMBUF_SZ is calculated for given cache_dir size Regards Chima
Re: [squid-users] coss configuration example
Hi Steven, I have tried following but it failed dd if=/dev/zero bs=1048576 count=2048 of=/cache_log/coss Part of Squid.conf -- cache_dir coss /cache_log/coss 2500 block-size=512 max-size=131072 cache_access_log /var/log/squid/access.log cache_log /var/log/squid/cache.log cache_store_log none cache_swap_log none Squid was compiled as ./configure --enable-storeio=coss --enable-coss-aio-ops --with-large-files While starting the squid i am getting error as 2006/08/14 19:36:37| WCCP Disabled. 2006/08/14 19:36:37| Ready to serve requests. 2006/08/14 19:36:37| COSS: /cache_log/coss: Rebuilding (0 % completed - 1/2500 stripes) 2006/08/14 19:36:39| COSS: /cache_log/coss: Rebuilding (5 % completed - 126/2500 stripes) 2006/08/14 19:36:40| COSS: /cache_log/coss: Rebuilding (10 % completed - 251/2500 stripes) 2006/08/14 19:36:42| COSS: /cache_log/coss: Rebuilding (15 % completed - 376/2500 stripes) 2006/08/14 19:36:43| COSS: /cache_log/coss: Rebuilding (20 % completed - 501/2500 stripes) 2006/08/14 19:36:44| COSS: /cache_log/coss: Rebuilding (25 % completed - 626/2500 stripes) 2006/08/14 19:36:45| COSS: /cache_log/coss: Rebuilding (30 % completed - 751/2500 stripes) 2006/08/14 19:36:47| COSS: /cache_log/coss: Rebuilding (35 % completed - 876/2500 stripes) 2006/08/14 19:36:48| COSS: /cache_log/coss: Rebuilding (40 % completed - 1001/2500 stripes) 2006/08/14 19:36:49| COSS: /cache_log/coss: Rebuilding (45 % completed - 1126/2500 stripes) 2006/08/14 19:36:50| COSS: /cache_log/coss: Rebuilding (50 % completed - 1251/2500 stripes) 2006/08/14 19:36:52| COSS: /cache_log/coss: Rebuilding (55 % completed - 1376/2500 stripes) 2006/08/14 19:36:53| COSS: /cache_log/coss: Rebuilding (60 % completed - 1501/2500 stripes) 2006/08/14 19:36:55| COSS: /cache_log/coss: Rebuilding (65 % completed - 1626/2500 stripes) 2006/08/14 19:36:57| COSS: /cache_log/coss: Rebuilding (70 % completed - 1751/2500 stripes) 2006/08/14 19:36:58| COSS: /cache_log/coss: Rebuilding (75 % completed - 1876/2500 stripes) 2006/08/14 19:36:59| COSS: /cache_log/coss: Rebuilding (80 % completed - 2001/2500 stripes) 2006/08/14 19:37:01| COSS: /cache_log/coss: Rebuilding (85 % completed - 2126/2500 stripes) 2006/08/14 19:37:02| COSS: /cache_log/coss: Rebuilding (90 % completed - 2251/2500 stripes) 2006/08/14 19:37:02| COSS: /cache_log/coss: Rebuilding (95 % completed - 2376/2500 stripes) FATAL: storeCossDirCloseTmpSwapLog: rename failed I don't know what went wrong. Regards Chima On 8/5/06, Steven Wilton <[EMAIL PROTECTED]> wrote: - Original Message ----- From: "chima s" <[EMAIL PROTECTED]> To: "squid mailing" Sent: Friday, August 04, 2006 8:24 PM Subject: [squid-users] coss configuration example > Hi, > > Can any one give the squid configuration to use coss cache_dir. > > My system info > > Fedora Core release 3 > Kernel : 2.6.10 > Squid : squid-2.6.STABLE2-20060804 > MEM: 2GB > HD : 2*72 GB SCSI > System Filesystem: reiserfs > > How COSS_MEMBUF_SZ is calculated for given cache_dir size > > Regards > Chima > The minimum config for a COSS dir is the following (for 100MB cache_dirs): cache_dir coss /var/spool/squid/coss 100 block-size=512 max-size=131072 or cache_dir coss /dev/sdb1 100 block-size=512 max-size=131072 If you set COSS to write to a flie, you must create the file manually. To do this, you can run: dd if=/dev/zero bs=1048576 count= of= If you set COSS to write direct to a block device, you need to make sure that the cache user has write permissions on the block device. The COSS_MEMBUF_SZ does not get set on a per-device basis. The magic number you need to be aware of is the block-size, which limits the size of the cache. The valid numbers are: block-size=512 - 8GB Max cache_dir size block-size=1024 - 16GB Max cache_dir size block-size=2048 - 32GB Max cache_dir size block-size=4096 - 64GB Max cache_dir size block-size=8192 - 128GB Max cache_dir size Steven
[squid-users] Web authentication password in transparent proxy
Hi, I am running the squid in transparent mode and also apache web server. In apache web server we have configured Web authentication (AccessFileName ..htaccess). If the request going through squid, its not asking for password and ant getting the page, but if i call the page without squid i am getting popup window which ask for usernamd and password. How do i get the web servers authentication window through transparent proxy. I read in FAQ for acceleartor mode to add #define AUTH_ON_ACCELERATION 1 in acl.c Will this help for my setup? Regards Chima
[squid-users] epoll performance
Hi, I am using FC2-Linux-2.6.10 Kernel in dual P4 Xeon Processer 3 GB Physical RAM 2 * 12 GB disked Cache_dir with different SCSI harddisk. OS File Systen - reiserfs Squid- 2.6.STABLE2 with cache_replacement_policy heap LFUDA memory_replacement_policy heap LFUDA Server is configured with in bridge mode and tproxy. Usually when there is 5 MB of traffic squid will use 80 to 90% of CPU (System load will be 2), after enabling the epoll it come down to 10 to 15% of CPU (System load will be 0.2). Then i added 5 more MB of traffic the CPU went to 40 to 50% (System load went to 3). What could be the problem as CPU usage and system load went high dramatically Regards Chima
[squid-users] coss configuration
Hi, I am trying to use coss cache_dir in squid-2.6.STABLE3 I craeted the cacge_dir dd if=/dev/zero bs=1048576 count=32768 of=/cache01/coss Squid.conf cache_dir coss /cache01/coss 32768 block-size=2048 max-size=131072 While starting squid i an getting the following error 2006/11/10 10:02:19| COSS block-size = 2048 bytes 2006/11/10 10:02:19| COSS largest file offset = 4194302 KB 2006/11/10 10:02:19| COSS cache_dir size = 33554432 KB FATAL: COSS cache_dir size exceeds largest offset Squid Cache (Version 2.6.STABLE3-20060916): Terminated abnormally. CPU Usage: 0.003 seconds = 0.001 user + 0.002 sys Maximum Resident Size: 0 KB Page faults with physical i/o: 1 Aborted What is there error means and how to handle this issue Regards Chima
[squid-users] COSS performance
Hi, I am running squid-2.6.Stable 3 on linux-2.6.10 kernel I am using squid epoll and coss cache_dir. System memory: 4 GB Cache Memory: 2 GB cache_dir coss /cache01/coss 1 block-size=2048 max-size=131072 cache_dir coss /cache02/coss 1 block-size=2048 max-size=131072 I am getting the below error message. storeCossCreateMemOnlyBuf: no free membufs. You may need to increase the value of membufs on the /cache01/coss cache_dir Do i require to change any configuration in squid? Regards Chima