Am Montag, 2. Februar 2009 13:30:13 schrieb Tobias Appel:
> On Mon, 2009-02-02 at 13:16 +0100, Michael Schwartzkopff wrote:
> > Am Montag, 2. Februar 2009 12:57:07 schrieb Tobias Appel:
> > (...)
> >
> > > It's 90% configured by hand through xml files. I would need a
> > > constraint that kills the resource group, demoted DRBD and promotes
> > > the other node to DRBD master and then start the resource group over
> > > there - but this seems to be not possible with heartbeat 2.1.4. I
> > > really tried every combination of constraint for pingd but it just does
> > > not work the way it has to be.
> > > Either it tries to promote the other node to DRBD master while the
> > > current node still runs all the resource and returns an error or it
> > > does nothing at all! It does not even stop the resource group.
> > >
> > > I've put one year work into this project and now I fail on the last and
> > > final step to making this HA cluster. I can't tell my boss that it will
> > > failover in 90% of the scenarios but not when the ethernet connection
> > > is down.
> >
> > hi,
> >
> > That setup is quite typical. I did this in a lot of szenarios and there
> > should not be any problems. If your send your CIB, perhaps we can help.
>
> I posted it in the IRC channel already, Andrew told me that my
> constraint is faulty - well I expected that.
> I then read the documentation from pacemaker, Andrew really did a
> wonderful job putting those pdf's together! There are a lot of examples
> for pingd but nothing worked out for me.
> HEre is my complete cib.xml: http://pastebin.com/d58a375a1
What I was seeing from your location constraint:
<rsc_location id="location_resources" rsc="ms_drbd">
<rule id="prefered_location_resources" score_attribute="pingd"
role="Master">
<expression attribute="pingd" id="66155d1f-2210-45ad-9010-
a14e48825ead" operation="defined"/>
</rule>
</rsc_location>
A node with no conenctivity still has pingd defined, but gets "0" points. If
you have a resource stickiness of 100 points, the DRBD has no reason to change
the node.
Since you have only one ping node make DRBD run on a node with pingd points:
<rsc_location id="ms_drbd:connected" rsc="ms_drbd">
<rule id="ms_drbd:connected:rule" score="-INFINITY" boolean_op="or">
<expression id="ms_drbd:connected:expr:undefined"
attribute="pingd" operation="not_defined"/>
<expression id="ms_drbd:connected:expr:zero"
attribute="pingd" operation="lte" value="0"/>
</rule>
</rsc_location>
in this case it doesn't matter if your DRBD is master or slave.
Please have a look about the real score calulation with the showscores script.
--
Dr. Michael Schwartzkopff
MultiNET Services GmbH
Addresse: Bretonischer Ring 7; 85630 Grasbrunn; Germany
Tel: +49 - 89 - 45 69 11 0
Fax: +49 - 89 - 45 69 11 21
mob: +49 - 174 - 343 28 75
mail: [email protected]
web: www.multinet.de
Sitz der Gesellschaft: 85630 Grasbrunn
Registergericht: Amtsgericht München HRB 114375
Geschäftsführer: Günter Jurgeneit, Hubert Martens
---
PGP Fingerprint: F919 3919 FF12 ED5A 2801 DEA6 AA77 57A4 EDD8 979B
Skype: misch42
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
