In HA control plane, each service has a dedicated container called "<service>_restart_bundle", which is run when the service configuration files change.
During a stack update / minor update converge, container <service>_restart_bundle will call pcs to restart the pacemaker resource associated with the service. E.g. for redis:
/usr/bin/bootstrap_host_exec redis /sbin/pcs resource restart --wait=600 redis_bundle
However, if - for whatever reason - the resource is in a state that prevent pacemaker from restarting it, pcs will complain and return an error:
2020-03-20T15:57:32.161477524+00:00 stdout F Fri Mar 20 15:57:32 UTC 2020: Restarting redis-bundle globally
2020-03-20T15:57:32.747605707+00:00 stderr F Error: Error performing operation: No such device or address
2020-03-20T15:57:32.747605707+00:00 stderr F redis-bundle is not running anywhere and so cannot be restarted
2020-03-20T15:57:32.747605707+00:00 stderr F
This causes the container to fail, and the stack update / minor update converge to fail as well.
Fix proposed to branch: master
Review: https:/