[Pacemaker] [SOLVED] Resource-Monitoring with an "On Fail"-Action

[Andrew Beekhof]

On Wed, Mar 31, 2010 at 1:16 PM, Tom Tux <tomtux80 at gmail.com> wrote:
> Hi
>
> From Novell-Support I got a PTF (Program Temporary Fix) which should
> handle this issue.
>
> I'm thinking, that the monitoring is working now. But I'm irritated
> with the output of the command "crm_mon -t1", which shows me the
> "last-rc-change" and the "last-run" of the monitor-operation. I have
> defined the monitor-operation for an certain resource every 10
> seconds, but the "last-run"-field of the "crm_mon -t1"-output doesn't
> change it's value. It changes it's value only, when he got no
> returncode with value "0" back and the failcount will be increased. Is
> this behaviour correct?

Short answer: Yes.
Long answer: The cluster does not want to know about every invocation
of the monitor and tells the lrmd to just tell us when something
changed.
There is significant bandwidth penalties for trying to keep track of
every single invocation so this is unlikely to change.

In retrospect, last-run is probably a bad name for that particular field.