I understand what the doc says about cache_mem, but it is the
only configurable I could see which might even remotely be
related.
We do get a fair amount of traffic through our proxy server -
about 200,000 to 300,000 hits per day.
I could handle it if it were only using 35MB of RAM - or even
100MB of RAM, but it seems to keep growing until it crashes.
We too have come to the conclusion that the only fix is to
kill and restart it every day or two. Seems a little extreme
for such a fast/reliable program. I have been very impressed
with what SQUID can do.
Regards,
Bill
Alex Rousskov wrote:
> On Thu, 4 Dec 1997, Bill Petersen wrote:
>
> > I am running Squid 1.1.18 I configure squid to
> >
> > cache_mem 8
> >
> > but find that over a week or so, squid uses over 100MB
> > of RAM. WHY?
>
> Read the documentation on "cache_mem" please. "cache_mem" does not specify
> the total memory requirement. "cache_mem" stands for the amount of RAM
> reserved for intransit and "hot" objects only.
>
> > Is there any way to limit how much RAM
> > squid will use?
>
> There is only one way, I guess: Limit your cache swap size. Release notes
> have a description on how to calculate Squid memory requirement given swap
> size.
>
> > Eventually it grows to around 200MB RAM
> > and crashes. I have 64MB of RAM in the system with 1GB of swap.
>
> 1GB swap / 10K avg_obj_size = 100K objects
> 100K objects * 100 bytes_per_hash_entry = 10MB RAM
>
> Thus, the hash table will cost you about 10MB of RAM. Other stuff will
> probably consume about 10MB more. Plus 8 MB for cache_mem. Thus, the total
> would be around 30-35MB. I am not sure why it grows to 200MB.
>
> Alex.
-- ------------------------------------------------------------------------ Bill Petersen email: brp@dsccc.com UNIX System, Web, Firewall, and Email Admin voice: 972-519-4249 Technical Services, DSC Communications Corp, Plano, Texas ------------------------------------------------------------------------ Communication, its one of those things we take for granted... ------------------------------------------------------------------------Received on Thu Dec 04 1997 - 17:08:50 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:37:51 MST