On Wed, Nov 5, 2008 at 02:19, Alex Strachan
<[EMAIL PROTECTED]> wrote:
>> -----Original Message-----
>> From: [EMAIL PROTECTED] [mailto:linux-ha-
>> [EMAIL PROTECTED] On Behalf Of Andrew Beekhof
>> Sent: Wednesday, 5 November 2008 2:29 AM
>> To: General Linux-HA mailing list
>> Subject: Re: [Linux-HA] resources unmanaged/managed - unexpected behavior
>>
>> Any chance you could attach the cib from just before you set
>> is-managed-default to true?
>>
>
> Unfortunately not.
>
> I added the --meta to the crm_resource command, and modified the order in
> which the resources are set to unmanaged/managed.  This is now performing
> well.

Ok, then the reason for the restart is that the resource definition
changed (the value of the is-managed "attribute" changed requiring a
restart).
By using --meta, you're telling the cluster that this is an option for
the PE - not the resource.  This puts it in a different namespace
which doesn't influence restarts.

Btw. In 1.0 you can put the whole cluster into maintenance mode by
setting the global maintenance-mode attribute.
This differs from is-managed in that it will also automatically cancel
any recurring monitor actions (preventing the cluster from even
noticing if anything fails).

Perhaps that helps.
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to