crash after stopping heartbeat

[Alan Robertson]

Stegbauer Thomas wrote:
> 
> should i take out the /dev/watchdog from /etc/ha.d/ha.cf? then
> 
> for what reason is it in then?
> 
> > you must be running watchdog, which is triggering the reboot.  we
> > finally
> > abandoned watchdog for that reason.
> 
> > > i am playing with heartbeat 0.4.7 on a suse linux 6.3 box and have the
> > problem
> > > when stopping the heartbeat:
> > >
> > > it stops all the resources, notifies at the end a heartbeat shutdown
> > complete.
> > >
> > > and after 60seconds? the machine makes a hard reboot??

How it's *supposed* to work is this...

If the machine hangs for 60 seconds, this will cause it to reboot.  The
device is supposed to be deactivated when heartbeat exits (and closes
/dev/watchdog).

And, in my tests, this is exactly how it works.  I have never seen it
shut the machine down in error.

However, I haven't used it in a few releases ;-)  I just reread the
shutdown code, and if it takes 60 seconds to shut down your resources
(long-running resources stop scripts), then this could happen.

However, if you're saying that 60 seconds after it said it was
completely shut down, that this happened, then I don't understand that.

	-- Alan Robertson
	   alnar@suse.com