Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations strongm on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

Squid freeze on OS/2

Status
Not open for further replies.

Max5Pier

MIS
May 21, 2003
21
CA
I've been running squid proxy-cache 2.4 on OS/2 server for a couple of months now and I'm realizing that the cache.log and acces.log are becoming bigger and bigger (150Mb and still counting).

I realized that the logs where getting so big Squid froze and none of the clients could access the internet. I tried to stop the proces ($ ps...) but it would not work, I had to reset the server. Since then I've experienced the freezes on many of the other servers and its becoming a real big problem...

Now I'm not sure that both events are related at all. I looked at the logs and all the entries look legitimate and it does not seem to explain the freezing...

If anybody as any ideas as to what is the parameter to limit the logs's size or why it freezes like that, please lat me know. Thanks.
 
Squid has a log rotate feature. You can set the number of rotations in the log_rotate parm of squid.conf, mine is at 10, and everynight in a scheduler i run "squid -k rotate". This will close all logs and rename them to xxx.log.0 where xxx is access, store and cache. Squid will then start a new set of empty logs. the next day the .0 will become .1 the current logs .0

I also have the hangs in os/2 and unable to stop squid with squid -k shutdown. a kill of the process always retuns a rc=162. The freezing of squid has nothing to do with the size of the logs. Before i started rotating my logs some got as big as 600 megs. At first I thought that squid was running out of file descriptors, 2.4 only has 256, so I implemented 2.5 stable 2 with at least 10000 handles. No luck. By the way I have a site that this freeze occurs daily. Out of 1600 sites, there are at least 9 a day that hang. The greater the load the worst it gets.

With 2.5 stable 2, when the error occurs, i am able to shutdown squid with either a ctrl-c in the squid window or a squid -k shutdown. I am unable to restart it though without a reboot of the server. The error indication i get, when i restart squid after the hang, is a warning error about a dns failure when squid opens the parent link. A nslookup of the same links works ok and squid's DNS test also works fine. Once I reboot the dns warning goes away. Almost sounds like an ip stack failure.

That's where i am today. I have to find a solution to this problem. We have at least 2 years to live with OS/2 before all the conversion to windows is complete.

The problem seems to have gotten worst since we implemented warp 4.5 convenience pack 2. There is alse more usage of squid since the customer is deploying more and more web applications.
That it for now. Hope this helps and puts you on another track. I will keep posting news as I test new things.
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top