On Wed, 12 Jul 2000, Felix Perlin wrote:
> I am receiving errors during "Hello World!" test.
> I am running Red Hat 6.1 with Apache and Squid
> 000.09| i-dflt 5079 1015.73 0 0.00 0 1
> 000.18| i-dflt 10199 1023.94 0 0.00 0 1
> 000.23| PortMgr.cc:44: error: 1/1 (98) Address already in use
> ...
Your system is probably not capable of sustaining 1000 req/sec load for
more than a few seconds. For example, you are probably running out of
ephemeral port space (you can use explicit port binding, but that is not
likely to help much because of the connections in TIME_WAIT state).
http://polygraph.ircache.net/doc/cltopts.html#--ports
If you need to simulate loads of that magnitude, you need to tune Linux
(or any other OS). Please look at our modifications for FreeBSD and try
to apply similar changes, where applicable.
http://polygraph.ircache.net/Tips/
I recall a similar posting from another Linux user; please see if you
can find it in the archives
http://polygraph.ircache.net/List/archive/
If you do not need 1000 req/sec loads, then you need to slow Robots
down. There are several ways to do it: use constant request rate
workload, add server side think time, etc.
> 000.23| Client.cc:337: error: 1/1623 (273) client ran out of new public oids
This is probably a separate issue, but I would fix the above problem
first and ignore oid errors for now.
Good luck,
Alex.
This archive was generated by hypermail 2b29 : Tue Jul 10 2001 - 12:00:14 MDT