[ClusterLabs] crm resource stop VirtualDomain - but VirtualDomain shutdown start some minutes later

[Ken Gaillot]

On Wed, 2022-02-16 at 18:09 +0100, Lentes, Bernd wrote:
> 
> ----- On Feb 16, 2022, at 12:52 AM, kgaillot kgaillot at redhat.com
> wrote:
> 
> > A transition is the set of actions that need to be taken in
> > response to
> > current conditions. A transition is aborted any time conditions
> > change
> > (here, the target-role being changed in the configuration), so that
> > a
> > new set of actions can be calculated.
> > 
> > Someone once defined a transition as an "action plan", and I'm
> > tempted
> > to use that instead. Plus maybe replace "aborted" with
> > "interrupted",
> > so then we'd have "Action plan interrupted" which is maybe a little
> > more understandable.
> 
> These "transition xxxx aborted" happen quite often.

Yes, they're quite normal. Basically they just mean that conditions
changed, so we need to check if anything needs to be done differently.

> 
> Feb 15 20:53:25 [15370] ha-idg-2       crmd:   notice:
> abort_transition_graph:  Transition 126 aborted by vm_documents-oo-
> meta_attributes-target-role doing modify target-role=Stopped:
> Configuration change | cib=7.27453.0 source=te_update
> _diff_v2:483
> path=/cib/configuration/resources/primitive[@id='vm_documents-
> oo']/meta_attributes[@id='vm_documents-oo-
> meta_attributes']/nvpair[@id='vm_documents-oo-meta_attributes-target-
> role'] complete=false
>                               
> Feb 15 20:53:00 [15370] ha-idg-2       crmd:     info:
> abort_transition_graph:  Transition 125 aborted by vm_amok-
> meta_attributes-target-role doing modify target-role=Stopped:
> Configuration change | cib=7.27452.0 source=te_update_diff_v2
> :483
> path=/cib/configuration/resources/primitive[@id='vm_amok']/meta_attri
> butes[@id='vm_amok-meta_attributes']/nvpair[@id='vm_amok-
> meta_attributes-target-role'] complete=true
> 
> Why is there sometimes "complete=true" and sometimes "complete=false"
> ?
> What does that mean ?
> 
> Bernd

"Complete" is whether all actions originally planned in the transition
were completed. For complete=true, the log is basically just a heads-up 
that the cluster needs to recheck things, since there's nothing to
actually abort.
-- 
Ken Gaillot <kgaillot at redhat.com>