On 02.04 19:49, Seewo Chen wrote: > > cache_replacement_policy heap LFUDA > > cache_swap_low 93 > > cache_swap_high 95 > > > > question: Why the cache swap don't reduce to 93% after reached 95%? > > What can I do under those case? it is probably because those numbers are too close to each other. You probably should use 95 and 90. Or 90 and 80. > >>You are using LFUDA to cache maximum size of objects > >>so that what is your minimum_object_size TAG > >>configuration? May be because of BIG sized object get > >>stored in cache so that cache_dir is not maintained to > >>the 93% to 95%. > > maximum_object_size 20480 KB > minimum_object_size is default , set to 0 KB > > now I do the following: > rm -rf /cache1 > rm -rf /cache2 not needed to remove whole cache, it should be enough to decrease cache size and let squid to sipe it by itself. > /etc/rc.d/init.d/squid -z > reboot > > How to prevent this case again? decrease cache size (9GB?) and make cache_swap_low and cache_swap_high differ a bit more. 2% is too low. Also, looking at it, you are running two caches on the same drive. it is just waste of resources. I recommend you creating one bigger partition and use one cache_dir. not 20GB, I'd try 18 (with cache_swap_* 95 and 90) or 19 (with cache_swap 90 and 80) GB. -- Matus UHLAR - fantomas, uhlar@xxxxxxxxxxx ; http://www.fantomas.sk/ Warning: I wish NOT to receive e-mail advertising to this address. Varovanie: na tuto adresu chcem NEDOSTAVAT akukolvek reklamnu postu. Depression is merely anger without enthusiasm.
