On Thu, 19 Aug 2004 [EMAIL PROTECTED] wrote:
By the time I went from DEV phase to PROD phase, my memory must have changed the meaning of the comments in squid.conf. I'm sorry. I thought it 'was' the max amount for hot objects. I had put 128MB on the developpement instance. the only reason I doubled it was to take the best out of the hardware. "the larger the better" ?.
Should I put it back to 8MB, or is 128 MB allright ?
You should only make it larger than the default if you know for certain there is plenty of free memory and the process size is not in danger of hitting an OS limit.
I forgot to tell you that all these crashes are seamless to the users : processes restart on their own : is that because only child processes core dump, and then primary squid process respawns new ones ?
It is because Squid automatically restarts itself on fatal errors.
What is said in your system messages log?
Regards Henrik
