LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: Again: LVS/Apache cluster freezes from time to time

To: "LinuxVirtualServer.org users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: Again: LVS/Apache cluster freezes from time to time
From: Jan Abraham <jan_abraham@xxxxxxx>
Date: Tue, 04 Nov 2003 19:00:14 +0100
Am Mo, 2003-11-03 um 18.54 schrieb Joseph Mack:
> are you running anything around the LVS, like Mon, HB...
> how are you replicating data between the DBs..

Nothing of that, but I've something new to report...

As I wrote, I've made the apache on one of the realservers listen to the
RIP and directed some traffic there. No freezes here.

On suggestion of Nicolas, I've checked the configuration of director and
realservers once again and adjusted several parameters to be nearly the
same as in Joseph's Howto. It didn't help, the realservers freezed.

So I did some more diagnostics and tried to look at the netstat output
during the freeze. I was a bit confused to see the number of ESTABLISHED
connections exploding from about 10 (normal operation) to 200 and more
during "freeze". So the server DOES accept connections, but Apache is
not able to process it for some reason.

Next result: I always asumed that something stopped me from starting new
processes during the freeze. This is wrong. Something stopps tasks that
accesses the /proc filesystem for process information. 
I can run "netstat -n", but I can't "netstat -n --program" when freezed.

I can do a "ls /proc/", but I can't do "ls /proc/<id>/" (<id> is the pid
of an apache thread)

That's strange, isn't it?
Jan


<Prev in Thread] Current Thread [Next in Thread>