RE: polygraph runs out of memory and quits

senthil wrote:
> Yes,this is on the client machine.
> and just as u have said the open sockets count increses and the
> response time is also high!! I'll try doing the no proxy test but
> I'm not sure how this is goin to prevent the client from running out
> of memory.
> 
> anyhow thanx for the quick response.
> 
> regards,
> Senthil

If the response times are high, you are almost certainly overloading
your cache - you need to reduce the request rate until response times
are acceptable (my organisation uses 2.5secs for misses, 0.2s for hits)
and memory is not exhausted. That is the maximum load which your cache
can withstand before it becomes unusable. 

A no-proxy test will prevent out-of-memory failures because these are
caused when polygraph's wait queues grow due to the fact that the proxy
is not responding to requests in a timely manner. If this continues
unchecked, the polyclt processes eventually exhaust all available memory
and exit. 

Doing a no-proxy test will reassure you that your test bed is capable of
these higher request rates - by eliminating the proxy, the queues do not
grow because they are being drained as replies arrive from the polysrv
processes. 

This has been covered several times on the polygraph-users list:
http://www.web-polygraph.org/mail-archive/users/200103/0002.html
http://www.web-polygraph.org/mail-archive/users/200103/0005.html
http://www.web-polygraph.org/mail-archive/users/200303/0005.html
http://www.web-polygraph.org/mail-archive/users/200303/0006.html

HTH,

Bryn.

--
Bryn M. Reeves
TCS Associate, Research & Development Team.
Avantis Limited, http://www.avantisworld.com/