With 8MB cache_mem and 5MB maximum_object_size_in_memory, any two large,
uncached items will go over the cache_mem level. Since you probably have
memory_pools on, squid will hang on to that extra memory in case it needs
it later (and it probably will in this config).
What effect does lowering maximum_object_size_in_memory to, say, 64k have?
-- Brian
On Thursday 09 August 2001 05:24 am, Jan Van der Veken wrote:
> I have a modest system running squid for a small lan.
> The box has the following specs:
> Pentium 233
> 64MB RAM, 128MB swap partition
> OS = Slackware Linux 7.1
> Squid version = 2.4.STABLE1
>
> The problem I'm seeing is that periodically, the memory
> usage of squid will go up but the allocated memory
> never gets released. After a while, I just have to
> shutdown and restart squid or it would eat all swap.
>
> I think I have traced it down to the use of streaming
> audio with windows media player by some users. It seems
> that whenever they are using this, the memory usage of
> squid increases linearly over time.
>
> I have the following parameters set in squid.conf.
> cache_mem 8 MB
> maximum_object_size 30240 KB
> maximum_object_size_in_memory 5000 KB
> cache_dir ufs /var/spool/squid/cache 300 16 256
Received on Thu Aug 09 2001 - 13:04:04 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:01:31 MST