[ClusterLabs] crm resource trace

[Ken Gaillot]

On Mon, 2022-10-17 at 12:43 +0200, Lentes, Bernd wrote:

The section you highlighted does contain the key difference:

> Oct 14 19:05:52 [26000] ha-idg-1    pengine:     info:
> rsc_action_digest_cmp:   Parameters to vm-genetrap_start_0 on ha-idg-
> 1 changed: was e2eeb4e5d1604535fabae9ce5407d685 vs. now
> 516b745764a83d26e0d73daf2c65ca38 (reload:3.0.14)
> 0:0;82:692:0:167bea02-e39a-4fbc-a09f-3ba4d704c4f9

vs

> Oct 14 19:26:33 [26000] ha-idg-1    pengine:     info:
> rsc_action_digest_cmp:   Parameters to vm-genetrap_monitor_30000 on
> ha-idg-1 changed: was 2c5e72e3ebb855036a484cb7e2823f92 vs. now
> d81c72a6c99d1a5c2defaa830fb82b23 (reschedule:3.0.14)
> 0:0;28:797:0:167bea02-e39a-4fbc-a09f-3ba4d704c4f9

In the first case, Pacemaker detected that the start action parameters
changed, but in the other case, it only detected that the recurring
monitor changed. Recurring monitors can be changed without requiring a
full restart.

I'm not sure why the start change wasn't detected in the second case.
Immediately after the log messages you showed for each case, there
should be a "saving inputs in <filename>" message. If you can privately
email me those two files, I can try to figure out what happened.
-- 
Ken Gaillot <kgaillot at redhat.com>