[Pacemaker] Ordering Constraint: Difference Between Stop and Migrate?

[David Vossel]

----- Original Message -----
> From: "Sven Arnold" <sven.arnold at localite.de>
> To: pacemaker at oss.clusterlabs.org
> Sent: Saturday, March 9, 2013 4:43:38 PM
> Subject: [Pacemaker] Ordering Constraint: Difference Between Stop and	Migrate?
> 
> Dear all,
> 
> Is it possible with ordering constraints to distinguish stopping a
> resource from migrating it to another node?

Take at look at this issue and the comment I posted.

http://bugs.clusterlabs.org/show_bug.cgi?id=5055#c3

I've looked into this in great detail. At the moment you have encountered a limitation of policy engine.  Hopefully someday we can fix this. The solution to this problem is unfortunately somewhat complex.

-- Vossel



> I think it is not possible since the resource manager would need to
> know
> the "migration downtime" in advance but anyways I will give it a try:
> 
> Given two resources A and B consider the following cases:
> 
> 1) A and B are stopped
> 
> When A is started, B is not affected
> When B is started, A has to be started before
> 
> 2) A and B are started
> 
> When A is stopped, B has to be stopped before
> When B is stopped, A is not affected
> 
> But:
> When A is migrating to another node (and thus stopped temporarily), B
> shall keep running.
> 
> The most appropriate rule would be:
> 
> order A_before_B 0: A B
> 
> but it would not help enforcing stopping B when shutting down A for
> more
> than a short period.
> 
> 
> The use case I have in mind is having a shared storage (DRBD, iSCSI
> Target) with configured failover on two nodes and an iSCSI Initiator
> and
> other services on other nodes. I would like to keep the dependent
> services running during a planned migration of the storage. And I
> would
> like to have a "correct" order configured when shutting down
> services.
> 
> Best regards,
> 
> Sven
> 
> _______________________________________________
> Pacemaker mailing list: 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://bugs.clusterlabs.org
>