>From: Richard Lyons <[EMAIL PROTECTED]> >Date: Mon, 1 Sep 2003 23:12:14 +1000 > >On Mon, 1 Sep 2003, Shpend Bakalli wrote: > >> (it continues to eat the mem and swaps), and it is not accounted in >> squid process (which grows up to 500 MB). When squid is shut down > >Forgive me for asking the obvious question: are you basing this >on the output of free? You're aware that the OS uses unallocated >pages as disk cache? > >Rick. >
I am aware, but I dont think that squid will start swaping if the OS is using 2 gigs of disk cache... the cache/buffer memory supposedly should be freed to the applications asking for it right? Output from Free... total used free shared buffers cached Mem: 3098532 3087732 10800 0 197972 1902248 -/+ buffers/cache: 987512 2111020 Swap: 2040244 19948 2020296 and after a while... total used free shared buffers cached Mem: 3098532 3087804 10728 0 171668 1798304 -/+ buffers/cache: 1117832 1980700 Swap: 2040244 43728 1996516 I have no other memory-eating processes in this machine, except the usual processes necessary for running squid... why its more important for kernel to keep the cache/buffer in-memory and swap squid out? can I alter this somehow? regards, Shpendi