Re: [squid-users] cache directory damaged, and mem usage

On 24 May 2001 00:26:58 +0200, Henrik Nordstrom wrote:
> Florin, if you want help you have to provide more detailed information.
>
> As said before, can you send the full HTML output of
>
> * cachemgr "General Runtime information"
> * cachemgr "Memory utilization"

It is here:

http://www.geocities.com/elf_too/squid-report.tar.gz

I also included 3 graphs in the tarball: CPU usage (cpu-day.png), HTTP
requests per second (hps-day.png) and memory usage (mem-day.png).
The memory usage is in megs, not in kilo (as it may seem from the
graph).

As you can see from the graphs, it doesn't grow too much if the load is
not too high.
As soon as the load grows beyond a certain limit, it starts to grow like
hell, over 1.5 GB or more! (the memory usage as reported by cachemgr.cgi
is not accurate, sometimes it's with 50% smaller than actual memory
usage)

> I want these two from three different times
>
> a) Just after startup and cache validation.

As you can see from the memory graph, this was done right after
restarting Squid, approximately at 00:15 AM.
See *-015.html files

> b) After having run steady with some load for a while, but without
> growing unexpectedly much. (if possible)

Done at 02:24 AM (so, it ran for 2 hours under light load before taking
this second snapshot; i hope it's enough)
See *-224.html files

> c) When the memory usage has grown unexpectedly

At 17:35
See *-1735.html files

Thank you.

-- 
Florin Andrei
"All operating systems suck.
Linux just sucks less" - an MIT guy