On Wednesday 21 April 2004 09:16, Martin Svensson wrote: > Hey all squid gurus! > > We have a cache to which 6000-7000 clients are > connecting to (2.5STABLE5). On a typical day we are > handling 70-100 requests/sec and 8-11Mbit in network > usage. > > The problem I am experiencing is that the machine > starts to swap (performance is still good though). We > have a cache of 90GB, 3GB of memory and cache_mem is > set to 128MB. We are caching quite large files to > speed up frequent downloads (up to 100MB files). > > Acccording to the FAQ, this setup should use: > 90GB cache = 900MB > cache_mem = +128MB > Add. mem = +20MB > Total recommended: 1048MB x 2 = 2096MB. > > I still have 1GB left of memory. No other memory > intense applications are running on this server and > still it swaps out as much as 600-700MB before I > manually restart the cache. It's running on RHEL 3.0. > > Any hints are appreciated. If more info is needed, > just let me know. I just don't see the reason to why > it swaps out so much.
/proc/meminfo, top, etc? -- vda
