[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-ha-dev
Subject:    Re: [Linux-ha-dev] heartbeat restart problem
From:       Alan Robertson <alanr () unix ! sh>
Date:       2001-02-25 18:43:10
[Download RAW message or body]

Juri Haberland wrote:
> 
> "Bene, Martin" wrote:
> >
> > Hi Alan,
> >
> > Just did some experiments with current heartbeat code (0.4.8l); the
> > combination of "nice_failback on" and "/etc/rc.d/init.d/heartbeat
> > restart" on a node currently holding resources gices unexpected
> > results:
> >
> > 1) currently owned resources are released
> > 2) heartbeat is restarted.
> >
> > Since restart is fast enough that the other side doesn't declare the
> > node dead, no failover takes place. On restart the local node finds
> > the other side to be up and running, so it doesn't try to re-aquire
> > the resources it previously released.
> >
> > Net result: we've got two nodes running heartbeat and talking happily
> > to each other while resources are sitting round unclaimed by any
> > node.

This shouldn't happen if you're running nice_failback on.  It might not take
over the resources, but it should squawk about it rather loudly and
frequently (like every second or so).  Or at least that's what I think...

> > Guess the easiest answer would be "don't do that then", but it's
> > still a bit unexpected.
> 
> Yep, I asked Alan to change the rc-script to do a real restart if called
> with "restart" because I needed and expected this behaviour. Before it
> would just reread the config file. It _does_ have the drawback that you
> expierenced. I don't have an idea how to prevent this other than "don't
> do it" but IMO it should do a restart on "restart" and only a reload on
> "reload" as most other rc-scripts do.

And, I just added a "long enough" sleep in the middle so that this should be
"much better".  Not perfect perhaps, but much better...

	-- Alan Robertson
	   alanr@unix.sh
_______________________________________________________
Linux-HA-Dev: Linux-HA-Dev@lists.community.tummy.com
http://lists.community.tummy.com/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/

[prev in list] [next in list] [prev in thread] [next in thread]