Ben Mckellar wrote:
> UP Time: 3822854.197 seconds
> Page faults with physical i/o: 299894571 <--- #### Is this a lot?
Probably.
It is in the range of a couple of tera-bytes of disk page faults, on a
Squid that has been running for 44 days with a very small request load..
> 2. Memory usage for squid via mallinfo():
> Total space in arena: 219853 KB
> Total in use: 224612 KB 102%
> Total free: 3469 KB 2% <--- #### That does not equal 100%
The % values are wrongly calculated. They are off by the amount of
holding blocks.
> I run a 60gig Squid Object Cache: Version 2.3.STABLE1 proxy server. I have
> a Watchguard firewall and i force everyone connecting on port 80 throught
> the proxy server on 3128.
>
> Proxy server specs:
> CPU p3 450mhz
> RAM 128
> HD 3 x 30gig scsi 10k rpm (2 are mounted as /cache1 and /cache2)
> Do I need more Ram in this machine perhaps? to fix the 2% Total free memory?
Not for the % which is not related to the amount of RAM installed in any
way.
An important fact from your cachemgr output (apart from the very high
amount of pagefaults) is that your Squid process is about twice as big
as the installed amount of RAM. I would recommend installing another
256MB of ram in that box (256MB for the Squid process, 128MB for the OS,
networking buffers and the rest) if you want it to perform well.
-- Henrik Nordstrom Squid hacker -- To unsubscribe, see http://www.squid-cache.org/mailing-lists.htmlReceived on Thu Feb 15 2001 - 15:33:38 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:58:01 MST