OpenBSD 3.5, Pound 1.6 built against openssl-0.9.7d.
Things work well for a while but eventually I get
these errors on the logs:
pound: MONITOR: worker exited on signal 11,
restarting...
I also run a script that runs curl to check that the
system is serving vaild results. It runs every five
minutes, and ocassionaly (once an hour or so) curl
returns the following:
curl: (52) Empty reply from server
I am running the same check through a different load
balancer and also directly against the machines and
never get this error.
Config:
#
ListenHTTP *,80
LogLevel 3
User XXX
Group XXX
Alive 5
Err500 "/usr/local/etc/error.html"
Err501 "/usr/local/etc/error.html"
Err503 "/usr/local/etc/error.html"
UrlGroup ".*"
BackEnd 10.0.0.101,8080,1
BackEnd 10.0.0.102,8080,1
#BackEnd 10.0.0.103,8080,1
BackEnd 10.0.0.104,8080,1
BackEnd 10.0.0.105,8080,1
BackEnd 10.0.0.106,8080,1
EndGroup
I have tried various OpenBSD tweaks to insure that
there was no resource exhaustion. I upped the limits
for the login class that pound runs under. I have
increased the following in the kernel:
maxusers,nmbclusters
I am almost positive that this is not a resource
exhaustion issue.
Any Ideas?
[...]
|