[Pacemaker] crm_resource -C vs. crm/resource/cleanup

[Andrew Beekhof]

On Tue, May 12, 2009 at 2:48 PM, Dejan Muhamedagic <dejanmm at fastmail.fm> wrote:
> Hi,
>
> On Mon, May 11, 2009 at 01:38:58PM -0400, Eliot Gable wrote:
>> I believe it is not expecting the resource to be running, but a
>> monitor action shows that it is running.
>
> If that was the case, then CRM would do something about it
> (stop/migrate).
>
>> Thus, it is not
>> technically in the correct state, so a warning should be
>> issued. If the resource is managed, then Pacemaker will start
>> it by itself, but expects it to be not running to begin with.
>
> Basically, the CRM should be the only one to start/stop the
> resources it manages. Now, perhaps after the resource cleanup
> (i.e. status section removed for the resource on some node), it
> could still reduce the severity. This warning I normally
> encounter on cluster startup. No idea how complex would that be.
> Andrew?

There's not really any way to know that we're in a post-cleanup scenario.
But for probes the log message could probably be downgraded to "info".

I'll do that once I recover my machine :-/