[Pacemaker] Looking for correct constraints

Michael Schwartzkopff misch at multinet.de
Fri Oct 30 19:51:38 UTC 2009


Am Freitag, 30. Oktober 2009 18:06:56 schrieb Lars Marowsky-Bree:
> On 2009-10-30T16:41:34, Andrew Beekhof <andrew at beekhof.net> wrote:
> > >> This does not make any sense. Sorry, let me try again.
> > >>
> > >> res1 start = set attribute
> > >> res1 stop = delete attribute
> > >>
> > >> rule for res2 with -inf score if attribute is set.
> > >
> > > better, just give all your nodes an attribute with the same value and
> > > use a colocation constraint with a score-attribute (IIRC)
> >
> > sorry, s/score-attribute/node-attribute/
>
> The problem with this obviously is that the attribute is not removed if
> res1 has been fenced.
>
> I still wonder what the use case is.
>
> If I have a) resA and resB, b) resB should only run if resA doesn't.
>
> So now resA fails. It'd be restarted elsewhere, negating the need for
> resB to run.
>
> What would cause resA to be unable to run anywhere else?

The customer.

Basically it is nagios with a database below. They have a split-site scenario 
and do not want to have a split-site cluster. So they have a database on every 
site. the nagios that writes the results to the local database, is preferred. 
If the local database is not available -> start the nagios, that writes to the 
remote database. Database replication is done seperately without the help of 
pacemaker.

Greetings,

-- 
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: misch at multinet.de
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




More information about the Pacemaker mailing list