[ClusterLabs] cluster okey but errors when tried to move resource - ?

[Ken Gaillot]

On Sat, 2023-06-03 at 15:09 +0200, lejeczek via Users wrote:
> Hi guys.
> 
> I've something which I'm new to entirely - cluster which is seemingly
> okey errors, fails to move a resource.

What pcs version are you using? I believe there was a move regression
in a recent push.

> I'd won't contaminate here just yet with long json cluster spits when
> fails but a snippet:
> 
> -> $ pcs resource move PGSQL-clone --promoted podnode1
> Error: cannot move resource 'PGSQL-clone'
>    1 <cib crm_feature_set="3.17.4" validate-with="pacemaker-3.6"
> epoch="8109" num_updates="0" admin_epoch="0" cib-last-written="Sat
> Jun  3 13:49:34 2023" update-origin="podnode2" update-
> client="cibadmin" have-quorum="1" update-user="root" dc-uuid="2">
>    2   <configuration>
>    3     <crm_config>
>    4       <cluster_property_set id="cib-bootstrap-options">
>    5         <nvpair id="cib-bootstrap-options-have-watchdog"
> name="have-watchdog" value="false"/>
>    6         <nvpair id="cib-bootstrap-options-dc-version" name="dc-
> version" value="2.1.6-1.el9-802a72226be"/>
>    7         <nvpair id="cib-bootstrap-options-cluster-
> infrastructure" name="cluster-infrastructure" value="corosync"/>
>    8         <nvpair id="cib-bootstrap-options-cluster-name"
> name="cluster-name" value="podnodes"/>
>    9         <nvpair id="cib-bootstrap-options-stonith-enabled"
> name="stonith-enabled" value="false"/>
>   10         <nvpair id="cib-bootstrap-options-last-lrm-refresh"
> name="last-lrm-refresh" value="1683293193"/>
>   11         <nvpair id="cib-bootstrap-options-maintenance-mode"
> name="maintenance-mode" value="false"/>
>   12       </cluster_property_set>
>   13       <cluster_property_set id="redis_replication">
>   14         <nvpair id="redis_replication-redis_REPL_INFO"
> name="redis_REPL_INFO" value="c8kubernode1"/>
>   15         <nvpair id="redis_replication-REDIS_REPL_INFO"
> name="REDIS_REPL_INFO" value="podnode3"/>
>   16       </cluster_property_set>
>   17     </crm_config>
> ...
> crm_resource: Error performing operation: Invalid configuration
> 
> This one line: (might be more)
> <nvpair id="redis_replication-redis_REPL_INFO" name="redis_REPL_INFO"
> value="c8kubernode1"/>
> puzzles me, as there is no such node/member in the cluster and so I
> try:

That's not a problem. Pacemaker allows "custom" values in both cluster
options and resource/action meta-attributes. I don't know whether redis
is actually using that or not.

> 
> -> $ pcs property unset redis_REPL_INFO --force
> Warning: Cannot remove property 'redis_REPL_INFO', it is not present
> in property set 'cib-bootstrap-options'

That's because the custom options are in their own
cluster_property_set. I believe pcs can only manage the options in the
cluster_property_set with id="cib-bootstrap-options", so you'd have to
use "pcs cluster edit" or crm_attribute to remove the custom ones.

> 
> Any & all suggestions on how to fix this are much appreciated.
> many thanks, L.
-- 
Ken Gaillot <kgaillot at redhat.com>