I'm running apache from woody which perl and php. Over the past few months, apache has occasionally starting eating up all the available memory, until it fills up the swap file and gets killed. I'm unable to trace this memory overflow to any particular event in the access or error logs. It doesn't even appear that anyone has requested a URL that runs a script around the time the overflow starts. It can take an hour or two until it's used up all the memory. I realize this is probably insufficient information to diagnose the problem, but I'm wondering: (1) if others have experienced memory overflow like this with apache 1.3.26-0woody3, and (2) what I might do to hunt down the cause? So far the only thing I've tried is a cron job which reports memory statistics every 20 minutes, and all I've discovered from that is that a certain process will sometimes expand indefinitely over a couple of hours. The system has 200M of RAM plus 186M of swap space, and gets on average a few HTTP requests per minute. There's no evidence of anything like a DoS. -- Adam Kessel http://bostoncoop.net/adam
Attachment:
pgpwtC2hUnjH7.pgp
Description: PGP signature