Re: 2.5 profiling

Grahame Bowland wrote:

 
> We ran squid HEAD on our main proxy server for a couple of days. We did
> discover that it grew in memory usage to approximately 460Mb instead of
> the normal 250Mb, and continued to grow until we restarted it.
>
> ./configure --enable-async-io=128 --enable-removal-policies
> --enable-poll --enable-underscores --enable-delay-pools --enable-snmp
>
> We do have some patches todo with the delay pools code, but there is no
> way they are causing a memory leak and 2.4-STABLE2 with the same patch
> set doesn't exhibit the problem.
>
> I'm more than happy to debug this if someone will give me hints how :)

For what it's worth, I didn't hit this while preparing for, or at, the
cacheoff I don't think. My Squid process was big (because of a high
cache_mem setting) but it didn't grow indefinitely--it flattened out
once the cache was full. I don't have an opinion at this point on
whether the process size was bigger than an equal 2.4 or 2.2STABLE5
Squid. I'll have to check that soon.

-- 
Joe Cooper <joe@swelltech.com>
http://www.swelltech.com
Web Caching Appliances and Support