On ons, 2008-05-28 at 09:36 +0200, [EMAIL PROTECTED] wrote: > Hello, I have a problem with Squid. Every day, over 0:00h (when logs are > rotated) my squid generates a core file that almost full all > '/var/spool/squid' partition. Looking at squid-cache logs, here is what it's > shown: > > 2008/05/27 00:06:12| helperOpenServers: Starting 40 'squirm' processes > 2008/05/27 00:06:12| ipcCreate: fork: (12) Cannot allocate memory
You need more swap, and probably also enable the sleep_after_fork directive. The UNIX way of starting a sub process (helper) is a bit odd, and can in worst case require a temporary memory resevation of 40 times the normal amount of memory for Squid while it is starting the helpers.. It's just a reservation, not acual use, but still needs swap space for the reservation to succeed. with sleep_after_fork enabled this can be limited to 2 times the normal amount.. Regards Henrik
