Hi all, Pacemaker supports the usual actions for OCF agents such as start, stop, and monitor.
Usually monitor is the only action configured as recurring. However it is possible to create a custom named action in an agent and configure it as recurring in Pacemaker. Currently, a recurring custom action that fails will be handled similarly to a failed monitor (taking the on-fail response, marking the resource as failed, etc.). That's not documented or tested, and I suspect it might be handled inconsistently. For Pacemaker 3.0.0, we're considering ignoring the results of custom recurring actions. They would still be run, they just wouldn't affect the state of the resource. With OCF_CHECK_LEVEL, it's possible to configure three different monitor behaviors, so I don't see much benefit to custom actions. What's your opinion? Does anyone use custom actions or find them potentially worthwhile? -- Ken Gaillot <kgail...@redhat.com> _______________________________________________ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/
