Re: [squid-users] Cache Size Limitation

Our cache filled up. Squid cosumed 99% of the CPU and filled it's memory
buffer, at which point it nearly filled our swap partition.
I tried to give it an abort, then core signal (kill -6), but for some
reason, it didn't write the core file.

Anyone care to share their thoughts on that one? It's really baffling me.
 Why would the process successfully abort, but not write a core? I know
it aborted because the CPU usage dropped back down to 0, after I sent
squid the -6 signal.

Very strange.

Regards,

Tim Rainier

Denis Vlasenko <vda@port.imtp.ilyichevsk.odessa.ua>
03/26/2004 08:42 AM
 
        To: trainier@kalsec.com, squid-users@squid-cache.org
        cc:
        Subject: Re: [squid-users] Cache Size Limitation

On Friday 26 March 2004 15:04, trainier@kalsec.com wrote:
> Is there a way to have squid automatically flush the cache when it
reaches
> it limit?
>
>
> I suppose I could, alternatively, write a shell script that runs every
> hour and analyses the size of the cache directory, then entities the
> swap.state file(s) and clears the directories when it gets close to the
> cache limit.
> I'd just rather use squid if it has the ability to do so.

Actually, the other way around.

At a certain point, it becomes easier to use modular tools
instead of trying to make one tool do everything imaginable.

If your cache dir lives in a dedicated partition, simple df
will tell you how large is your cache now.

BTW, why do you need it?

-- 
vda