Hi,

we've got strange problem. we set very busy site on pound, it works
fine for about 10 minutes and after that our monitoring system sends
us SMS that it can't connect to :80 port. we try to do telnet to
pound:80 but got:
trying... telnet can't connect.

At the same time when we do ps auwx | grep pound we can see pound
process. nothing appears in to log after monitoring's alert, last line
is:
(at)4000000046c44f981df6094c 1.2.3.4 GET /file.jpg HTTP/1.1 - HTTP/1.1
200 OK (www.site.com/- -> 192.168.0.10:80) 0.003 sec

This setup uses daemontools to run new pound process when actual dies.
But in this situation it doesn't start new because one is running.

We try to kill -9 POUND_PID but it doesn't help - pound still exists
as process on system.
We do kill -9 POUND_PID several times but nothing happens.
Then we do: kill -9 supervise_PID, then daemontools runs new
superservse and it runs new pound process. Now when we do ps auwx |
grep pound we can see two processes of pound.

Is it known issues? What can we do to debug this situation and to
resolve our problem?

Several weeks ago we've got similar problem, but that was not so
production site so we forgot about it. Now problem returns.

pound.cfg:
User            "pound"
Group           "pound"
Daemon          0
LogLevel        2
LogFacility     -
Alive           30
Client          60
Control         "/var/run/pound.socket"

ListenHTTP
        Address IP
        Port    80
        xHTTP   0
End

ListenHTTPS
        Address IP
        Port    443
        xHTTP   0
        Cert    "/usr/local/etc/pound/cert.pem"
End

Service
        Emergency
                Address 192.168.0.22
                Port    80
        End
        BackEnd
                Address 192.168.0.10
                Port    80
        End
        BackEnd
                Address 192.168.0.27
                Port    80
        End
End

proxy# pound -V
16/Aug/2007 15:32:41 +0200: starting...
Version 2.3
  Configuration switches:
    --disable-super
    --with-owner=pound
    --with-group=pound
Exiting...


OS: FreeBSD proxy2.site.com 6.1-RELEASE-p10 FreeBSD 6.1-RELEASE-p10
#0: Wed Nov 29 10:18:34 CET 2006
root(at)proxy2.site.com:/usr/obj/usr/src/sys/PROXY  i386

Any ideas?
TIA,
p.

We're facing the same problem here.

The interesting thing is that it only happens if you have active connections.

We have two Pounds in a CARP fashion: the first one is always active
(primary), the secondary is on only if has a failure with the primary.

If I try to send a KILL (or a HUP) to pound process in the secondary
server all goes well. However if I try to send a KILL (or a HUP) in
the main server nothing happens.

I have to restart the machine so the process can read the new pound.cfg.

Im using FreeBSD 6.1 too but Pound version is 2.2.

Is that a know issue ?

Thnx,

c0re
[...]

> We try to kill -9 POUND_PID but it doesn't help - pound still exists[...]

<snip>
[...]

What's the state of the process, as reported by top / ps axuww?  I've 
seen python processes go unkillable, and I believe its related to this bug:

http://groups.google.com/group/mailing.freebsd.bugs/browse_thread/thread/16915d2170956df/8b840e6909b76a93%238b840e6909b76a93

My experience was on FreeBSD 6.2 - I'm unsure if the patch ever made it 
to 6.x, though I assume so since the issue is a few years old by now.

Regards,[...]

Hi,

changes with:

peceka wrote:[...]

didn't help. pounds with high load still hangs and don't want to go hell. ;) :(
[...]

I've got this.
[...]

We've got pound 2.3, is this variable in this version?
[...]

Changing kill'ing style from -9 to -QUIT didn't help.
[...]

I think that this isn't a problem... But we will look for this.

Thanks,
p.