[squid-users] SO_FAIL in store_log under high load
Hi, I experience high cache miss due to swap out failure under high load. Background setup: --- I run squid 2.6 stable 4-20061009 on Redhat AS server version 4 update 3 (64 bits kernel) The server hardware is dell 2950. 4 GB Ram, 6 hard drives, each 164 GB. Virtual memory is set to 20 GB. I compile the squid with --enable-async-io=10, --enable-follow-x-forwarded-for, --enable-auth=basic, --disable-wccp, --enable-snmp, --enable-x-accelerator-vary, --enable-remove-policies=lru I configure 5 cache_dir, each has 140 GB on each hard drive, 50 GB is dedicated for caches. The logs are write to one separate hard drive. Cache_mem is set to 2.5 GB I setup squid to run in reverse proxy mode with 100 domains, each domain has an upstream origin server. The first 33 domains share one origin server, the second 33 domains share the second origin server, the last 34 domain share the third origin server (total of 3 origin servers) disable the log on access_log. Two type of object sizes: 220 KB and 2KB. 50% request for 220KB and 50% request for 2 KB I prepopulate all the objects in the cache server (20 objects)10 objects of 220KB, 10 objects of 2KB (2000 objects each domain). Each object has time to live 15 minutes Behavior: Randomly accessing 20 objects: --Under low load, approximate 60 r/s, the server operates well with cpu at 30-60%, request hit rate, bytes hit rate is approximate 95% --As soon as I increase the load to higher than this, there is a lot of SO_FAIL in the store.log (swap failure) and thus the object is not valid anymore and the server starts to retrieve objects from the origin server even though it should not (object has not changed). The request/byte hit rate goes down to 50%. My questions are 1) what has caused the swap failure? Is it due to async threads or disk failure? 2) what is the optimum thread value that should one choose when compile under async-io? 3) Is there any other parameters in the squid.conf might impact the ability to swap an object out to disk? Note, for a few first runs, I was able to achive 600r/s, but now the best I can do is 200r/s with a lot of SO_FAIL and the request/byte hit hover over 50%. I even try to restart with empty cache, clear all cache_dir but that does not fix it. Hope there is someone has insight about this or have had some experience on this issue can give me some advices, suggestions to improve the operation of my cache. Thanks, Khanh
Re: [squid-users] SO_FAIL in store_log under high load
fre 2007-01-19 klockan 16:45 -0500 skrev Nguyen, Khanh, INFOT: 1) what has caused the swap failure? Is it due to async threads or disk failure? Anything in cache.log? 2) what is the optimum thread value that should one choose when compile under async-io? Depends a bit on the hardware. You have to experiment a bit. Having too many whill cause the I/O queues to get too large, to few will make Squid think the drive is overloaded before it is.. 3) Is there any other parameters in the squid.conf might impact the ability to swap an object out to disk? Not much. Regards Henrik signature.asc Description: Detta är en digitalt signerad meddelandedel
RE: [squid-users] SO_FAIL in store_log under high load
In cache.log has some queue congestion prior to SO_FAIL appearance and during SO_FAIL appearance: 2007/01/18 14:08:27| squidaio_queue_request: WARNING - Queue congestion 2007/01/18 14:08:30| squidaio_queue_request: WARNING - Queue congestion 2007/01/18 14:08:34| squidaio_queue_request: WARNING - Queue congestion 2007/01/18 14:08:39| squidaio_queue_request: WARNING - Queue congestion Should I try to lower the thread number to see if that yields less SO_FAIL? Is there anything that I can try to improve the situation? Thanks, Khanh -Original Message- From: Henrik Nordstrom [mailto:[EMAIL PROTECTED] Sent: Friday, January 19, 2007 5:11 PM To: Nguyen, Khanh, INFOT Cc: squid-users@squid-cache.org Subject: Re: [squid-users] SO_FAIL in store_log under high load fre 2007-01-19 klockan 16:45 -0500 skrev Nguyen, Khanh, INFOT: 1) what has caused the swap failure? Is it due to async threads or disk failure? Anything in cache.log? 2) what is the optimum thread value that should one choose when compile under async-io? Depends a bit on the hardware. You have to experiment a bit. Having too many whill cause the I/O queues to get too large, to few will make Squid think the drive is overloaded before it is.. 3) Is there any other parameters in the squid.conf might impact the ability to swap an object out to disk? Not much. Regards Henrik
RE: [squid-users] SO_FAIL in store_log under high load
fre 2007-01-19 klockan 18:02 -0500 skrev Nguyen, Khanh, INFOT: In cache.log has some queue congestion prior to SO_FAIL appearance and during SO_FAIL appearance: 2007/01/18 14:08:27| squidaio_queue_request: WARNING - Queue congestion No other warnings? Should I try to lower the thread number to see if that yields less SO_FAIL? The above message could be a sign that there is too many threads, but in itself it's a harmless warning. The swapout failure on the other hand is a sign there maybe is too few threads, or that I/O is overloaded. If you see swapout failures and no other warnings then the problem is probably in the load average calculation for the cache dir gets above 1000. See fs/aufs/store_dir_aufs.c storeAufsDirCheckLoadAv Regards Henrik signature.asc Description: Detta är en digitalt signerad meddelandedel
RE: [squid-users] SO_Fail -1 in store.log
What is the best approach to upgrade to 2.5.STABLE7 from 2.5.STABLE6? I see info on how to apply patches but not a new STABLE release. Just install the new version ontop of the old and restart Squid. Same as when patching except that there is no doubt in what Squid version you are using. Is this really the only option? I can't fix STABLE7? 2.5.STABLE7 is 2.5.STABLE6 + all patches and a few other cosmetic changes to bring everything together in a nice package. Going from one 2.5.STABLE to another 2.5.STABLE or applying a patch to a 2.5.STABLE is all the same effort (reinstall ontop of existing installation, then restart Squid), and by using the current STABLE release you can be confident that your questions will be taken seriously when you have trouble with stability or to reach some site. I tried to install 2.5STABLE7 but after doing .configure, make and make install I run squid -v and still STABLE6. I checked to config.log file and these look to be the only errors: configure:8051: checking Default FD_SETSIZE value configure:8081: gcc -o conftest -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include-g conftest.c -lrt -lpthread -lm -lresolv -lbsd -lnsl 15 configure:8073: warning: return type defaults to `int' configure: In function `main': configure:8076: warning: implicit declaration of function `exit' configure:8100: checking Maximum number of filedescriptors we can open configure:8172: gcc -o conftest -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include-g conftest.c -lrt -lpthread -lm -lresolv -lbsd -lnsl 15 configure:8119: warning: return type defaults to `int' configure: In function `main': configure:8167: warning: implicit declaration of function `exit' configure:8199: checking Default UDP send buffer size configure:8225: gcc -o conftest -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include-g conftest.c -lrt -lpthread -lm -lresolv -lbsd -lnsl 15 configure:8212: warning: return type defaults to `int' configure:8244: checking Default UDP receive buffer size configure:8270: gcc -o conftest -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include-g conftest.c -lrt -lpthread -lm -lresolv -lbsd -lnsl 15 configure:8257: warning: return type defaults to `int' configure:8289: checking Default TCP send buffer size configure:8315: gcc -o conftest -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include-g conftest.c -lrt -lpthread -lm -lresolv -lbsd -lnsl 15 configure:8302: warning: return type defaults to `int' configure:8338: checking Default TCP receive buffer size configure:8364: gcc -o conftest -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include-g conftest.c -lrt -lpthread -lm -lresolv -lbsd -lnsl 15 configure:8351: warning: return type defaults to `int' configure:8386: checking if sys_errlist is already defined configure:8420: checking for libresolv _dns_ttl_ hack configure:8429: gcc -o conftest -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include-g conftest.c -lrt -lpthread -lm -lresolv -lbsd -lnsl 15 /tmp/ccSM7OfG.o(.text+0x7): In function `main': /root/squid-2.5.STABLE7/configure:8425: undefined reference to `_dns_ttl_' collect2: ld returned 1 exit status configure: failed program was: #line 8422 configure #include confdefs.h extern int _dns_ttl_; int main() { return _dns_ttl_; ; return 0; } configure:8445: checking if inet_ntoa() actually works configure:8469: gcc -o conftest -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include-g conftest.c -lrt -lpthread -lm -lresolv -lbsd -lnsl 15 configure:8458: warning: return type defaults to `int' configure:8495: checking for working statvfs() interface configure:8514: gcc -c -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include conftest.c 15 configure:8534: checking for _res.nsaddr_list Is are these warnings preventing the upgrade? What else could I check? Thanks.
RE: [squid-users] SO_Fail -1 in store.log
What is the best approach to upgrade to 2.5.STABLE7 from 2.5.STABLE6? I see info on how to apply patches but not a new STABLE release. Just install the new version ontop of the old and restart Squid. Same as when patching except that there is no doubt in what Squid version you are using. Is this really the only option? I can't fix STABLE7? 2.5.STABLE7 is 2.5.STABLE6 + all patches and a few other cosmetic changes to bring everything together in a nice package. Going from one 2.5.STABLE to another 2.5.STABLE or applying a patch to a 2.5.STABLE is all the same effort (reinstall ontop of existing installation, then restart Squid), and by using the current STABLE release you can be confident that your questions will be taken seriously when you have trouble with stability or to reach some site. I tried to install 2.5STABLE7 but after doing .configure, make and make install I run squid -v and still STABLE6. I checked to config.log file and these look to be the only errors: configure:8051: checking Default FD_SETSIZE value configure:8081: gcc -o conftest -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include-g conftest.c -lrt -lpthread -lm -lresolv -lbsd -lnsl 15 configure:8073: warning: return type defaults to `int' configure: In function `main': configure:8076: warning: implicit declaration of function `exit' configure:8100: checking Maximum number of filedescriptors we can open configure:8172: gcc -o conftest -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include-g conftest.c -lrt -lpthread -lm -lresolv -lbsd -lnsl 15 configure:8119: warning: return type defaults to `int' configure: In function `main': configure:8167: warning: implicit declaration of function `exit' configure:8199: checking Default UDP send buffer size configure:8225: gcc -o conftest -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include-g conftest.c -lrt -lpthread -lm -lresolv -lbsd -lnsl 15 configure:8212: warning: return type defaults to `int' configure:8244: checking Default UDP receive buffer size configure:8270: gcc -o conftest -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include-g conftest.c -lrt -lpthread -lm -lresolv -lbsd -lnsl 15 configure:8257: warning: return type defaults to `int' configure:8289: checking Default TCP send buffer size configure:8315: gcc -o conftest -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include-g conftest.c -lrt -lpthread -lm -lresolv -lbsd -lnsl 15 configure:8302: warning: return type defaults to `int' configure:8338: checking Default TCP receive buffer size configure:8364: gcc -o conftest -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include-g conftest.c -lrt -lpthread -lm -lresolv -lbsd -lnsl 15 configure:8351: warning: return type defaults to `int' configure:8386: checking if sys_errlist is already defined configure:8420: checking for libresolv _dns_ttl_ hack configure:8429: gcc -o conftest -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include-g conftest.c -lrt -lpthread -lm -lresolv -lbsd -lnsl 15 /tmp/ccSM7OfG.o(.text+0x7): In function `main': /root/squid-2.5.STABLE7/configure:8425: undefined reference to `_dns_ttl_' collect2: ld returned 1 exit status configure: failed program was: #line 8422 configure #include confdefs.h extern int _dns_ttl_; int main() { return _dns_ttl_; ; return 0; } configure:8445: checking if inet_ntoa() actually works configure:8469: gcc -o conftest -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include-g conftest.c -lrt -lpthread -lm -lresolv -lbsd -lnsl 15 configure:8458: warning: return type defaults to `int' configure:8495: checking for working statvfs() interface configure:8514: gcc -c -g -O2 -Wall -D_REENTRANT -I/usr/kerberos/include conftest.c 15 configure:8534: checking for _res.nsaddr_list Is are these warnings preventing the upgrade? No,configure has it's tests which can fail or not, for instance. What else could I check? Thanks. - check whether squid was build ; should be in relative to your unpacked distribution root : `updist`/src/squid (so to speak) Check then for instance : `updist`/src/squid -v `updist` = dir which contains configure If that's all OK, check the log/output of 'make install' : check for any errors. M.
RE: [squid-users] SO_Fail -1 in store.log
- check whether squid was build ; should be in relative to your unpacked distribution root : `updist`/src/squid (so to speak) This is there. So far so good. Check then for instance : `updist`/src/squid -v This returned STABLE7. Still good. But squid -v with out /src/ returns STABELE6 `updist` = dir which contains configure That is there. If that's all OK, check the log/output of 'make install' : check for any errors. During make install I get a lot of: Nothing to be done for. Making all in ip_user make[3]: Entering directory `/root/squid-2.5.STABLE7/helpers/external_acl/ip_user' make[3]: Nothing to be done for `all'. make[3]: Leaving directory `/root/squid-2.5.STABLE7/helpers/external_acl/ip_user' Making all in ldap_group make[3]: Entering directory `/root/squid-2.5.STABLE7/helpers/external_acl/ldap_group' make[3]: Nothing to be done for `all'. make[3]: Leaving directory `/root/squid-2.5.STABLE7/helpers/external_acl/ldap_group' Making all in unix_group make[3]: Entering directory `/root/squid-2.5.STABLE7/helpers/external_acl/unix_group' make[3]: Nothing to be done for `all'. make[3]: Leaving directory `/root/squid-2.5.STABLE7/helpers/external_acl/unix_group' Making all in wbinfo_group make[3]: Entering directory `/root/squid-2.5.STABLE7/helpers/external_acl/wbinfo_group' make[3]: Nothing to be done for `all'. make[3]: Leaving directory `/root/squid-2.5.STABLE7/helpers/external_acl/wbinfo_group' Making all in winbind_group make[3]: Entering directory `/root/squid-2.5.STABLE7/helpers/external_acl/winbind_group' make[3]: Nothing to be done for `all'. make[3]: Leaving directory `/root/squid-2.5.STABLE7/helpers/external_acl/winbind_group' make[3]: Entering directory `/root/squid-2.5.STABLE7/helpers/external_acl' make[3]: Nothing to be done for `all-am'. make[3]: Leaving directory `/root/squid-2.5.STABLE7/helpers/external_acl' make[2]: Leaving directory `/root/squid-2.5.STABLE7/helpers/external_acl' make[2]: Entering directory `/root/squid-2.5.STABLE7/helpers' make[2]: Nothing to be done for `all-am'. make[2]: Leaving directory `/root/squid-2.5.STABLE7/helpers' make[1]: Leaving directory `/root/squid-2.5.STABLE7/helpers' make[1]: Entering directory `/root/squid-2.5.STABLE7' make[1]: Nothing to be done for `all-am'. make[1]: Leaving directory `/root/squid-2.5.STABLE7'
RE: [squid-users] SO_Fail -1 in store.log
On Wed, 9 Feb 2005, Brad Taylor wrote: I tried to install 2.5STABLE7 but after doing .configure, make and make install I run squid -v and still STABLE6. I checked to config.log file and these look to be the only errors: How did you install 2.5.STABLE6? You need to install 2.5.STABLE7 in the exact same manner. Regards Henrik
RE: [squid-users] SO_Fail -1 in store.log
... ... `updist`/src/squid -v This returned STABLE7. Still good. But squid -v with out /src/ returns STABELE6 Well hard to say, but that simply means that the new binary, didn't go to where it was supposed to be by make install. As stated by others, make sure that this installation uses the same parameters as the previous one (configure 'prefix' ; for instance). M.
[squid-users] SO_Fail -1 in store.log
I've searched the archives and looked in the definitive guide but can't find any solutions to SO_Fail -1 for all image files I see in my store.log file. I know that this means squid could not save the object to disk. I'm trying to just cache images but my access log is showing TCP_MISS for gif files. I believe the cause is SO_Fail but I don't know the fix for SO_Fail. My cache log looks clean.
RE: [squid-users] SO_Fail -1 in store.log
I've searched the archives and looked in the definitive guide but can't find any solutions to SO_Fail -1 for all image files I see in my store.log file. I know that this means squid could not save the object to disk. I'm trying to just cache images but my access log is showing TCP_MISS for gif files. I believe the cause is SO_Fail but I don't know the fix for SO_Fail. My cache log looks clean. You need to use the latest stable release (2.5.STABLE7) to avoid this issue. M.
RE: [squid-users] SO_Fail -1 in store.log
You need to use the latest stable release (2.5.STABLE7) to avoid this issue. What is the best approach to upgrade to 2.5.STABLE7 from 2.5.STABLE6? I see info on how to apply patches but not a new STABLE release. Is this really the only option? I can't fix STABLE7? B
RE: [squid-users] SO_Fail -1 in store.log
On Mon, 7 Feb 2005, Brad Taylor wrote: What is the best approach to upgrade to 2.5.STABLE7 from 2.5.STABLE6? I see info on how to apply patches but not a new STABLE release. Just install the new version ontop of the old and restart Squid. Same as when patching except that there is no doubt in what Squid version you are using. Is this really the only option? I can't fix STABLE7? 2.5.STABLE7 is 2.5.STABLE6 + all patches and a few other cosmetic changes to bring everything together in a nice package. Going from one 2.5.STABLE to another 2.5.STABLE or applying a patch to a 2.5.STABLE is all the same effort (reinstall ontop of existing installation, then restart Squid), and by using the current STABLE release you can be confident that your questions will be taken seriously when you have trouble with stability or to reach some site. Regards Henrik
[squid-users] SO_FAIL
Hello Squid does not save internetfiles on the harddisk. In the store.log, i see the following messages: 1092911664.677 SO_FAIL -1 AA5601EB2B243693AFEAFAEF68C230BF 200 1092911664 1092815319-1 text/html 15488/13312 GET http://www.qmail.org/ 1092911664.731 SO_FAIL -1 AA5601EB2B243693AFEAFAEF68C230BF 200 1092911664 1092815319-1 text/html 15488/14760 GET http://www.qmail.org/ 1092911664.781 SO_FAIL -1 AA5601EB2B243693AFEAFAEF68C230BF 200 1092911664 1092815319-1 text/html 15488/15488 GET http://www.qmail.org/ 1092911664.781 SO_FAIL -1 AA5601EB2B243693AFEAFAEF68C230BF 200 1092911664 1092815319-1 text/html 15488/15488 GET http://www.qmail.org/ 1092911664.821 SO_FAIL -1 AA5601EB2B243693AFEAFAEF68C230BF 200 1092911664 1092815319-1 text/html 15488/15488 GET http://www.qmail.org/ 1092911665.507 SO_FAIL -1 E92BF1F9FA38BCA0CB53FF7B795CA79C 200 1092911665 1047680620-1 image/gif 36/36 GET http://www.qmail.org/red.gif 1092911665.553 SO_FAIL -1 6294F0BE30FE84B743E3B80BD9F6CF3A 200 1092911665 978674019-1 image/gif 1444/1444 GET http://www.qmail.org/powered-by-djbdns.gif My system is a Debian woody with kernel 2.4.26, Squidversion is: squid1:~# /opt/squid/sbin/squid -v Squid Cache: Version 2.5.STABLE6 configure options: --prefix=/opt/squid-2.5.STABLE6 --exec-prefix=/opt/squid-2.5.STABLE6 --enable-auth=ntlm,basic --enable-external-acl-helpers=wbinfo_group --with-samba-sources=/usr/src/samba-3.0.4 squid1:~# I have not found dokumentation to this messages. Can anyone help me ? Tino
RE: [squid-users] SO_FAIL
Hi Can you give a brief copy and paste of the access.log And too your conf file. Kind Regards Brent Clark -Original Message- From: Glatzel Tino [mailto:[EMAIL PROTECTED] Sent: Thursday, August 19, 2004 12:54 PM To: '[EMAIL PROTECTED]' Subject: [squid-users] SO_FAIL Hello Squid does not save internetfiles on the harddisk. In the store.log, i see the following messages: 1092911664.677 SO_FAIL -1 AA5601EB2B243693AFEAFAEF68C230BF 200 1092911664 1092815319-1 text/html 15488/13312 GET http://www.qmail.org/ 1092911664.731 SO_FAIL -1 AA5601EB2B243693AFEAFAEF68C230BF 200 1092911664 1092815319-1 text/html 15488/14760 GET http://www.qmail.org/ 1092911664.781 SO_FAIL -1 AA5601EB2B243693AFEAFAEF68C230BF 200 1092911664 1092815319-1 text/html 15488/15488 GET http://www.qmail.org/ 1092911664.781 SO_FAIL -1 AA5601EB2B243693AFEAFAEF68C230BF 200 1092911664 1092815319-1 text/html 15488/15488 GET http://www.qmail.org/ 1092911664.821 SO_FAIL -1 AA5601EB2B243693AFEAFAEF68C230BF 200 1092911664 1092815319-1 text/html 15488/15488 GET http://www.qmail.org/ 1092911665.507 SO_FAIL -1 E92BF1F9FA38BCA0CB53FF7B795CA79C 200 1092911665 1047680620-1 image/gif 36/36 GET http://www.qmail.org/red.gif 1092911665.553 SO_FAIL -1 6294F0BE30FE84B743E3B80BD9F6CF3A 200 1092911665 978674019-1 image/gif 1444/1444 GET http://www.qmail.org/powered-by-djbdns.gif My system is a Debian woody with kernel 2.4.26, Squidversion is: squid1:~# /opt/squid/sbin/squid -v Squid Cache: Version 2.5.STABLE6 configure options: --prefix=/opt/squid-2.5.STABLE6 --exec-prefix=/opt/squid-2.5.STABLE6 --enable-auth=ntlm,basic --enable-external-acl-helpers=wbinfo_group --with-samba-sources=/usr/src/samba-3.0.4 squid1:~# I have not found dokumentation to this messages. Can anyone help me ? Tino
Re: [squid-users] SO_FAIL
On Thu, 19 Aug 2004, Glatzel Tino wrote: Hello Squid does not save internetfiles on the harddisk. In the store.log, i see the following messages: 1092911664.677 SO_FAIL -1 AA5601EB2B243693AFEAFAEF68C230BF 200 1092911664 1092815319-1 text/html 15488/13312 GET My system is a Debian woody with kernel 2.4.26, Squidversion is: squid1:~# /opt/squid/sbin/squid -v Squid Cache: Version 2.5.STABLE6 Probably this: http://www.squid-cache.org/Versions/v2/2.5/bugs/#squid-2.5.STABLE6-ufs_no_valid_dir Regards Henrik
Re: [squid-users] SO_FAIL
Quoting Henrik Nordstrom [EMAIL PROTECTED]: On Wed, 7 Jul 2004, Sunil Mohan Ranta wrote: i am getting SO_FAIL messages in store.log and no cache is being created on my system the size of of cache_dir is constant at 17MB Any error messages in cache.log? REgards Henrik Strangely enough I have been having the same problem (no corresponding error in cache.log BTW), so I backed up my squid.conf, deleted all the entries in my cache dir and used the default squid.conf, modifying only the following: http_port 80 http_port 3128 http_port 8080 cache_dir ufs /share/cache 100 16 256 http_access allow all cache_effective_user squid cache_effective_group squid And the problem dissapeared! I haven't gone so far as to track down what part of my old config was causing the problem, but at least now I know it was something in the config (and not a problem with how squid was compiled or a directory permissions problem). I'll post an update when I have traced the cause. Regards, Rob Hadfield
[squid-users] How I fixed it... Re: [squid-users] SO_FAIL
Quoting [EMAIL PROTECTED]: Quoting Henrik Nordstrom [EMAIL PROTECTED]: On Wed, 7 Jul 2004, Sunil Mohan Ranta wrote: i am getting SO_FAIL messages in store.log and no cache is being created on my system the size of of cache_dir is constant at 17MB Any error messages in cache.log? REgards Henrik Strangely enough I have been having the same problem (no corresponding error in cache.log BTW), so I backed up my squid.conf, deleted all the entries in my cache dir and used the default squid.conf, modifying only the following: http_port 80 http_port 3128 http_port 8080 cache_dir ufs /share/cache 100 16 256 http_access allow all cache_effective_user squid cache_effective_group squid And the problem dissapeared! I haven't gone so far as to track down what part of my old config was causing the problem, but at least now I know it was something in the config (and not a problem with how squid was compiled or a directory permissions problem). I'll post an update when I have traced the cause. Regards, Rob Hadfield Ok, I replicated the problem by: * Changing: cache_replacement_policy heap LFUDA * Stopped Squid * Deleted the contents of my cache_dir * Started Squid (cache dir rebuilt itself) * store.log started logging the following for everything: ... SO_FAIL -1 ... * Stopped Squid * Deleted the contents of my cache_dir * Changed: cache_replacement_policy lru * Started Squid (cache dir rebuilt itself) * store.log started logging the following swapouts: ... SWAPOUT 00 ... ... SWAPOUT 00 0001 ... etc ... Now this may be due to my misunderstanding of how cache_replacement_policy heap LFUDA is supposed to work, but the comments in the config say this about it: # The heap LFUDA policy keeps popular objects in cache regardless of # their size and thus optimizes byte hit rate at the expense of # hit rate since one large, popular object will prevent many # smaller, slightly less popular objects from being cached. So is it my misunderstanding, or is it a bug? Regards, Rob Hadfield
Re: [squid-users] How I fixed it... Re: [squid-users] SO_FAIL
Now this may be due to my misunderstanding of how cache_replacement_policy heap LFUDA is supposed to work, but the comments in the config say this about it: # The heap LFUDA policy keeps popular objects in cache regardless of # their size and thus optimizes byte hit rate at the expense of # hit rate since one large, popular object will prevent many # smaller, slightly less popular objects from being cached. So is it my misunderstanding, or is it a bug? LFUDA is supposed to work yes, but your Squid needs to be build with support for the heap policies (compile time option). Any output from squid -k parse? Regards Henrik
Re: [squid-users] How I fixed it... Re: [squid-users] SO_FAIL
Quoting Henrik Nordstrom [EMAIL PROTECTED]: Now this may be due to my misunderstanding of how cache_replacement_policy heap LFUDA is supposed to work, but the comments in the config say this about it: # The heap LFUDA policy keeps popular objects in cache regardless of # their size and thus optimizes byte hit rate at the expense of # hit rate since one large, popular object will prevent many # smaller, slightly less popular objects from being cached. So is it my misunderstanding, or is it a bug? LFUDA is supposed to work yes, but your Squid needs to be build with support for the heap policies (compile time option). Any output from squid -k parse? Regards Henrik Ahh ok, my mistake - I did not configure-compile with --enable-heap-replacement or --enable-removal-policies=list of policies No output from squid -k parse. Regards, Rob Hadfield
Re: [squid-users] SO_FAIL
On Wed, 7 Jul 2004, Sunil Mohan Ranta wrote: i am getting SO_FAIL messages in store.log and no cache is being created on my system the size of of cache_dir is constant at 17MB Any error messages in cache.log? REgards Henrik
Re: [squid-users] SO_FAIL
actuall i had stable1 and it use to work fine. i upgrated to stable5, and got this problem It seems that I have the same or a very similar problem: No problems using squid-2.5.STABLE4-20040120 and before. Object are written on disk and logged with SWAPOUT in store.log. Using squid-2.5.STABLE5-20040609 or squid-2.5.STABLE6-20040709 there is always an SO_FAIL entry in store.log and no cache objects are written to disk. Permissions are not the problem. Debugging shows that Squid calculated the dirno to -1 (see storeSwapOutStart): 2004/07/09 15:36:46| storeSwapOut: http://ta-nea.dolnet.gr/data/D2004/D0709/2el10c.gif 2004/07/09 15:36:46| storeSwapOut: store_status = STORE_PENDING 2004/07/09 15:36:46| storeSwapOut: mem-inmem_lo = 0 2004/07/09 15:36:46| storeSwapOut: mem-inmem_hi = 115992 2004/07/09 15:36:46| storeSwapOut: swapout.queue_offset = 0 2004/07/09 15:36:46| storeSwapOut: lowest_offset = 105568 2004/07/09 15:36:46| storeSwapOut: swapout_size = 115992 2004/07/09 15:36:46| storeSwapOutStart: Begin SwapOut 'http://ta-nea.dolnet.gr/data/D2004/D0709/2el10c.gif' to dirno -1, fileno 2004/07/09 15:36:46| storeSwapMetaBuild: http://ta-nea.dolnet.gr/data/D2004/D0709/2el10c.gif 2004/07/09 15:36:46| storeCreate: no valid swapdirs for this object There is only one cache_dir entry in the config file: cache_dir ufs /home/squid/cache 25000 100 256 The configuration file, configure parameters, compiler and operating system are exactly the same. As I can't found a solution, I made a rollback to STABLE4. Operating System: IRIX 6.5.22 System: SGI Origin 200 Compiler: MIPSpro Compilers: Version 7.41 Heiko Heiko Schlichting| Freie Universität Berlin [EMAIL PROTECTED] | Zentraleinrichtung für Datenverarbeitung (ZEDAT) Telefon +49 30 838-54327 | Fabeckstraße 32 Telefax +49 30 838454327 | 14195 Berlin
Re: [squid-users] SO_FAIL
help please no related links found in google problem details system : fc2 squid : stable squid/2.5.STABLE5-20040629 lots of free space, actuall i had stable1 and it use to work fine. i upgrated to stable5, and got this problem cache_dir is owned by squid:squid is has ug+rwx tail store.log 1089299522.640 RELEASE -1 352AD5007DFA372DE85A728349B2A43D 200 1089299522-1-1 AIM/HTTP 0/0 POST http://XXX/data? 1089299522.900 SO_FAIL -1 44C39A9F3B8C67E5659B700A1BDF3E0B 200 1089299438 1001921864-1 image/gif 1155/1155 GET http://XXX/image/gifhome_3.gif 1089299522.918 SO_FAIL -1 207928A36772E9456E727F3A9DF51BF0 200 1089299438 1001921865-1 image/gif 1133/1133 GET http://XXX/image/gifhome_5.gif 1089299523.036 SO_FAIL -1 57D34087E74C0F9394388BAA1B26BE8D 200 1089299438 1001921867-1 image/gif 1072/1072 GET http://XXX/image/gifhome_7.gif 1089299523.052 SO_FAIL -1 BB61DCB65D8A53C6D584530B75D6DA4E 200 1089299438 1001921868-1 image/gif 1138/1138 GET http://XXX/gifhome_9.gif all of the entries are either SO_FAIL (mainly) or RELEASE tail access.log 1089299631.917564 XXX TCP_MISS/200 294 GET http://XXX/tab-separator.gif - DIRECT/63.210.164.48 image/gif 1089299632.157604 XXX TCP_MISS/200 404 GET http://XXX/refresh.gif - DIRECT/63.210.164.46 image/gif 1089299632.276762 XXX TCP_MISS/200 354 POST http://XXX/notify/ - DIRECT/216.155.194.191 text/plain almost all of the entries are TCP_MISS/DIRECT the squid.conf is almost default one, with very few changes. help please On Wed, 07 Jul 2004 23:50:27 +0530, Sunil Mohan Ranta [EMAIL PROTECTED] wrote: i am getting SO_FAIL messages in store.log and no cache is being created on my system the size of of cache_dir is constant at 17MB how do i enable caching, and get rid of this SO_FAIL message ?? help please !! -- killall -9 killall http://students.iiit.net/~sunilmohan/
RE: [squid-users] SO_FAIL
problem details system : fc2 squid : stable squid/2.5.STABLE5-20040629 lots of free space, actuall i had stable1 and it use to work fine. i upgrated to stable5, and got this problem cache_dir is owned by squid:squid is has ug+rwx tail store.log 1089299522.640 RELEASE -1 352AD5007DFA372DE85A728349B2A43D 200 1089299522-1-1 AIM/HTTP 0/0 POST http://XXX/data? 1089299522.900 SO_FAIL -1 Do you have additional error info in cache.log ? M.
[squid-users] SO_FAIL
i am getting SO_FAIL messages in store.log and no cache is being created on my system the size of of cache_dir is constant at 17MB how do i enable caching, and get rid of this SO_FAIL message ?? help please !! -- killall -9 killall http://students.iiit.net/~sunilmohan/
Re: [squid-users] SO_FAIL
My first suggestion would be to look at the permissions of the cache directory. Does the squid user have write permissions to it? Regards, Tim Rainier Sunil Mohan Ranta [EMAIL PROTECTED] 07/07/2004 02:20 PM To: [EMAIL PROTECTED] cc: Subject:[squid-users] SO_FAIL i am getting SO_FAIL messages in store.log and no cache is being created on my system the size of of cache_dir is constant at 17MB how do i enable caching, and get rid of this SO_FAIL message ?? help please !! -- killall -9 killall http://students.iiit.net/~sunilmohan/
RE: [squid-users] SO_FAIL
i am getting SO_FAIL messages in store.log and no cache is being created on my system the size of of cache_dir is constant at 17MB - Check squid's cache.log for more info. - Check your system logs too, watchout for system messages concerning for instance disk space, disk I/O, others... M.