topic Re: how to stop node auto reboot after hearbeat failure? . in Operating System - HP-UX

how to stop node auto reboot after hearbeat failure? .

somanaboina_522 — Mon, 12 Aug 2013 13:23:25 GMT

Dear All,

I have a 2 node cluster of hp-ux 11i v3 os servers, few days back there was network issue in my datacenter.

Due to hearbeat failure of the cluster one of the node from the cluster got auto reboot.

here my question is how to disable this option of node auto-reboot in case of hearbeat failure.

do i need to install any extra packages for this feature.

Thanks&Regards

somana

Re: how to stop node auto reboot after hearbeat failure? .

Patrick Wallek — Tue, 13 Aug 2013 13:31:11 GMT

As far as I know you cannot stop the auto reboot. The auto-reboot is a function of SG itself. It is designed to prevent what is known as a "split-brain" syndrome. "Split-brain syndrome" is a situation where you lose communication between the cluster nodes and the cluster reforms on BOTH nodes, which is NOT a good thing.

Basically the process is this (very simplified):

1) Heartbeat communication is lost

2) The cluster nodes compete to take control of the lock disk

3) The winner of the lock disk takes over the cluster and the loser of the lock disk reboots.

So, basically you DO NOT want to turn off the auto-reboot function. Your cluster behaved exactly as it should have in this situation.

Re: how to stop node auto reboot after hearbeat failure? .

Stephen Doud — Tue, 13 Aug 2013 18:57:08 GMT

Short of reducing the cluster to single-node operation or providing a switchless HB network (cross-over cable), there is no way to avoid a node TOC if all HB networks fail.

Re: how to stop node auto reboot after hearbeat failure? .

somanaboina_522 — Wed, 14 Aug 2013 06:21:03 GMT

Thank you patrick...