Try copy paste output result of squidclient mgr:info The possible 1. out of file descriptor 2. low average median response time of cache miss, cache hit. 3. Your cache_dir is fully utilized, so squid does many release and swapout object. On Wed, Jul 23, 2008 at 10:43 AM, Marcos Dutra <macdutra@xxxxxxxxx> wrote: > Hi people > > I have a server with redhat enterprise 5 Xeon 3.06 4 processors, 8Gb > RAM and 160GB SAS disk, well I ran the command in linux "netstat -an > |grep 3128| wc" and when arrives in 2500 connections, it is very slow. > How can I improve performance, I changed from diskd to aufs and I > don't get any performance, I changed RPM binary to compiled from the > source too. > Detais, i would like put in this server about 3000 - 4000 users. > > Thanks for advice > Marcos >
