Hi
> Squid Version: Squid-2.0/PATCH
> Clock Speed: 300 Mhz
> SCSI Disk Size: 17 Gb (the partition where squid is installed has 5 Gb)
> RAM: 128 Mb
> 56 DialUp user can simultaneously access squid
>
> cache_mem 48
This is definitely a problem. Your cache is going to store almost half
of your on-disk store in RAM... it may seem like that is going to make
things nice and speedy, but with only 128Mb of ram, you will find that
the machine swaps. When it swaps, it slows the whole of Squid down.
run 'vmstat 1' and watch the 'si' and 'so' values: I bet that they are
significant.
Since you are running lots of other services, your will find that
there are other big processes running too.
> cache_dir /usr/local/squid/cache 100 16 256
> What do you think about this configuration ? What do you suggest me ?
Change 'cache_mem' to 1 and shutdown and restart Squid.
Also: try and find the biggest process on the machine and restart it
occasionally (this is probably named or innd.) In the worst case, add
ram until the machine starts swapping... :)
Oskar
Received on Fri Apr 09 1999 - 07:04:31 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:45:46 MST