Hi Dough, you didn't say anything about the HA version you're using. I don't know if I can help you, but I got the following problems with HAv2 under heavy load which probably fit your situation:
a) Load raises. b) Monitor action on a resource times out because of heavy load. c) HA tries to stop resource as it don't know the state of the resource any more. d) Stop action times out because of heavy load. e) Node running the resource is fenced. Action: Raise the timeout values for monitor (and stop) actions. There have been some discusions on this list about that topic. Best regards Andreas Mock > -----Ursprüngliche Nachricht----- > Von: "Doug Eubanks" <[EMAIL PROTECTED]> > Gesendet: 21.08.08 18:14:38 > An: [email protected] > Betreff: [Linux-HA] Server reboots as soon as load average hits exactly 30 > We are using HA & DRBD for two Zimbra servers. > > If a server becomes overloaded (begins receiving a lot of e-mail all at > once), the load average begins to climb. It takes a few minutes, but > eventually it reaches exactly 30 and in that instant, the server reboots > (the default HA action) and the process begins to start on the other server > (which takes several minutes to start). If the load begins to climb on that > server because of a large queue that's backedup, it reboots and the process > falls back to the other server. I've seen it take 4 or 5 of these failovers > before it gets "caught up" and stops. > > I'm trying to track down what's rebooting the server and how best to handle > it. > > 1. HA may be rebooting the server because it missed a heartbeat (I doubt > this because it happens when the load average reaches exactly 30) > 2. HA may be rebooting the server because of the load, does HA do this > and how do I configure it not to do this if it does? > 3. Lastly, I may be able to configure Zimbra to reject incoming > connections after a certain load, but this isn't the ideal situation for > our > case. > > Can someone help shed some light on this? I would appreciate any comments > or suggestions! > > Thanks! > > Doug Eubanks > _______________________________________________ > Linux-HA mailing list > [email protected] > http://lists.linux-ha.org/mailman/listinfo/linux-ha > See also: http://linux-ha.org/ReportingProblems > _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
