[Pacemaker] Bug in ptest -s for promotion scores?

Michael Schwartzkopff misch at clusterbau.com
Wed Jan 19 04:37:14 EST 2011


On Wednesday 19 January 2011 09:52:46 Andrew Beekhof wrote:
> On Wed, Jan 19, 2011 at 9:37 AM, Michael Schwartzkopff
> 
> <misch at clusterbau.com> wrote:
> > Hi,
> > 
> > I have pacemaker-1.0.10 on opensuse11.3 from the clusterlabs repo
> > installed.
> > 
> > When I have a ms DRBD resource with a filesystem depending on the master
> > state of the DRBD and look at the scores I get:
> > 
> > # ptest -sL
> > (...)
> > resDRBD:0 promotion score on hakurs27: 10020
> > resDRBD:1 promotion score on hakurs28: 10000
> > resDRBD:0 promotion score on hakurs27: 1000000
> > resDRBD:1 promotion score on hakurs28: -1000000
> > (...)
> > 
> > please note that lines 1 and 3 are identical. also lines 2 and 4 a re
> > identical. So does resDRBD:0 get 10020 points or inf? And what would be
> > the promotion score of resDRBD:0 on node hakurs28?
> > 
> > Anybody with a good explanation?
> 
> For technical reasons the placement algorithm in 1.0 needs to be
> re-executed under some conditions (I forget exactly which).  What
> you're see is the "scores" from that recomputation.  The new scores
> are +/-infinity because we've already been through it once.
> 
> Basically, the first scores are the "real" ones.
> 
> Pretty sure I removed this from 1.1

hi,

I moves to opensuse's pacemaker-1.1.2. Same result:

# ptest -sL
(...)
resDRBD:1 promotion score on hakurs07: 10000
resDRBD:0 promotion score on hakurs08: 10000
resDRBD:1 promotion score on hakurs07: INFINITY
resDRBD:0 promotion score on hakurs08: -INFINITY

shouldn't it be

resDRBD:1 ... hakurs07
resDRBD:0 ... hakurs07
resDRBD:1 ... hakurs08
resDRBD:0 ... hakurs08

?

-- 
Dr. Michael Schwartzkopff
Guardinistr. 63
81375 München

Tel: (0163) 172 50 98
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
URL: <https://lists.clusterlabs.org/pipermail/pacemaker/attachments/20110119/428d63eb/attachment-0003.sig>


More information about the Pacemaker mailing list