NOSRV and retries

Hey

I'm running benchmarks on apache -> haproxy -> thin/rails, to try and get a configuration working.

The problem I am having is with 'option redispatch' and 'retries'.

I will set the following

option redispatch
retries 3
contimeout 5000

Then running ab to benchmark. When it reaches the 5 sec contimeout rather than retrying it will say NOSRV and sQ in the log, and return a 502 error code.

I'm using ver 1.3.15.2, in tcp mode, balance roundrobin and maxconn 1 on each thin/rails instance.

I'm confused as to why its returning 502 errors rather than redispatching.

Can anyone help?

http://clk.atdmt.com/UKM/go/msnnkmgl0010000007ukm/direct/01/ Received on 2008/06/29 16:55