Re: [squid-users] unlined used 99% CPU

A common problem that will cause this is configureing your cache size to be
the same as available HD space, but good practice says you should leave more
space than is needed available. And cache should be on it's own partition,
so that it's not competing with anything like /var/log and /tmp for the free
space.

With my cache on it's own partition my rule of thumb is the partition should
be at least 15% larger than the specified cache size.

If it's too late to separate it out, I would say that you need to pad in for
space needed by other file systems sharing the partition, so the numbers will
vary depending on what you have installed, and how much space is free on the
drive, how often your logs are rotated, etc...

On Thursday 21 April 2005 11:26 am, Mr. Pine wrote:
> Hi,
> I'm using squid. The problem is that when my cache
> reach to 50% full it try to run unlinke.in this case
> it takes 3 min and got 99% of CPU. after a while it
> crashed.
> I don'nt know if there is a misconfiguration or bug
> ????
> TNX for your notic
>
> here is my configuration:
> CPU: Intel(R) Pentium(R) 4 CPU 3.00GHz
> RAM: 2 GB
> ===
> Squid Cache: Version 2.5.STABLE5
> Linux Fedora core 3 (2.6.10)
> squid configuration:
> =====
> cache_mem 100 MB
> cache_dir ufs /proxy_cache/cache1 2400 16 256
> cache_dir ufs /proxy_cache/cache2 2400 16 256
> cache_dir ufs /proxy_cache/cache3 2400 16 256
> cache_dir ufs /proxy_cache/cache4 2400 16 256
> cache_dir ufs /proxy_cache/cache5 2400 16 256
> =====
>
> __________________________________________________
> Do You Yahoo!?
> Tired of spam? Yahoo! Mail has the best spam protection around
> http://mail.yahoo.com

-- 
Michael Wray
AimConnect, an S4F Inc. Company
918.524.1010 ext 106
mwray@aimconnect.com
http://www.aimconnect.com