RE: [squid-users] DNS round robin / httpd_accel, again...

Thanks - That's what I suspected; using ab to run request tests with high
connection numbers and high concurrency confirms that it distributes it
almost exactly 50% (~1% diff in my really informal test).

ab -k -n 60 -t 60 -c 16 http://nodes.example.com/ ends up hitting both nodes
about equally (based upon access log for each server, after staring with an
empty log):
node1:/var/log/apache# wc -l access.log
   1174 access.log
node2:/var/log/apache# wc -l access.log
   1190 access.log

It seems to be working very well. :)

Sean

-----Original Message-----
From: Henrik Nordstrom [mailto:hno@hem.passagen.se]
Sent: Tuesday, June 05, 2001 11:25 PM
To: sean.upton@uniontrib.com
Cc: squid-users@squid-cache.org
Subject: Re: [squid-users] DNS round robin / httpd_accel, again...

A single browser will probably give a uneven distribution as it is
connections that are round-robin not requests. Under higher loads this
should even out. Open connections are also re-used in a round-robin
fashion.

--
Henrik Nordstrom
Squid Hacker
sean.upton@uniontrib.com wrote:
> 
> I think I fixed the problem by changing my configuration so that I used
the
> defaults for ipcahce_size, fqdncache_size, and positive_dns_ttl; I
> previously had 0, becuase I was under the mistaken impression that I
wanted
> to NOT cache IPs.  Now,  it seems to work, though it seems that the load
is
> unevenly distributed (at least under the light load that my single browser
> is providing).  I am under the assumption that under a heavier (normal)
> stress load that this will even out?