[Pacemaker] Group not started/stopped in correct order with -INF collocation

Kulovits Christian - OS ITSC Christian.Kulovits at austrian.com
Mon Jul 18 04:38:20 EDT 2011


Hi Andrew,

But whats about the -Inf collocation. The groups must not be active on the same node. This implies an order in shutting down and starting up, isn´t it? On the other hand, if both groups gets started on different nodes there should not be any order dependency since startup of one group may last a long time and the other should not wait that time.

Regards, Christian

From: Andrew Beekhof [mailto:andrew at beekhof.net]
Sent: Dienstag, 05. Juli 2011 03:46
To: The Pacemaker cluster resource manager
Subject: Re: [Pacemaker] Group not started/stopped in correct order with -INF collocation


On Fri, Jul 1, 2011 at 7:23 PM, Kulovits Christian - OS ITSC <Christian.Kulovits at austrian.com<mailto:Christian.Kulovits at austrian.com>> wrote:
Hello Andrew,

No, there are no ordering constraint between the groups!

Then how is pacemaker supposed to know there is a start/stop order that should be observed?

Regards, Christian

From: Andrew Beekhof [mailto:andrew at beekhof.net<mailto:andrew at beekhof.net>]
Sent: Freitag, 01. Juli 2011 09:31

To: The Pacemaker cluster resource manager
Subject: Re: [Pacemaker] Group not started/stopped in correct order with -INF collocation


On Tue, May 31, 2011 at 10:05 PM, Kulovits Christian - OS ITSC <Christian.Kulovits at austrian.com<mailto:Christian.Kulovits at austrian.com>> wrote:
Hello,
In a 2-node cluster two groups (prod, test) are defined with a collocation score="-INFINITY". When the node running the group prod failed, the test group on the other node was stopped and the prod group was started. After the failed node was online again, the prod group was moved to the other node. BUT the test group was first started, then the prod group was stopped and started on the other node. There was a very short time where both groups were active on the same node.

Do you have an ordering constraint between them?

We use a RA to set the home directory for some users, and so the path gets first set and then unset by this sequence. There is no order defined between these groups to enable parallel starts on different nodes.
Is this a definition problem or a bug. Nevertheless this behavior looks strange for me!

Kind regards, Christian

Mit freundlichen Grüßen / with best regards
Christian Kulovits
____________________________________________

[cid:image001.jpg at 01CC4536.CFE31CA0]
AUSTRIAN AIRLINES
Christian Kulovits
ITSOC Central System & Database Services
Senior IT System Engineer
Head Office
Office Park 2, P.O. Box 100
1300 Vienna-Airport, Austria

*   Phone:     +43 (0)5 1766   11557<tel:%2B43%20%280%295%201766%C2%A0%C2%A0%2011557>
Ê   Fax:         +43 (0)5 1766 511557<tel:%2B43%20%280%295%201766%20511557>
È   Mobile:     +43 (0)664 80111 11557<tel:%2B43%20%280%29664%2080111%2011557>
*   email:      christian.kulovits at austrian.com<mailto:christian.kulovits at austrian.com>
ý   www:       www.austrian.com<http://www.austrian.com/>
____________________________________________


________________________________________________

Austrian Airlines AG, Office Park 2, P.O. Box 100, 1300 Vienna-Airport, Austria, registered office: Vienna, registered with Vienna Commercial Court under FN 111000k, DVR 0091740. This e-mail is confidential and is subject to disclaimers. Details can be found at: http://www.austrian.com/disclaimer.



_______________________________________________
Pacemaker mailing list: Pacemaker at oss.clusterlabs.org<mailto:Pacemaker at oss.clusterlabs.org>
http://oss.clusterlabs.org/mailman/listinfo/pacemaker

Project Home: http://www.clusterlabs.org
Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
Bugs: http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker


_______________________________________________
Pacemaker mailing list: Pacemaker at oss.clusterlabs.org<mailto:Pacemaker at oss.clusterlabs.org>
http://oss.clusterlabs.org/mailman/listinfo/pacemaker

Project Home: http://www.clusterlabs.org
Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
Bugs: http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clusterlabs.org/pipermail/pacemaker/attachments/20110718/826e1380/attachment-0003.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.jpg
Type: image/jpeg
Size: 6923 bytes
Desc: image001.jpg
URL: <https://lists.clusterlabs.org/pipermail/pacemaker/attachments/20110718/826e1380/attachment-0003.jpg>


More information about the Pacemaker mailing list