[ClusterLabs] Is this "fatal exit" behavior too strict for an invalid value of the stonith-watchdog-timeout property?

[Xin Liang]

Hi,

I have observed that setting the stonith-watchdog-timeout property to a
positive number causes Pacemaker to exit immediately if sbd.service is not
active. Additionally, when sbd is active, Pacemaker will exit if
stonith-watchdog-timeout is set to a value less than the SBD watchdog
timeout.

My question is: Is this "fatal exit" behavior too strict for an invalid
value of the stonith-watchdog-timeout property? Could we instead display a
"Configuration invalid" error, similar to the behavior when setting
stonith-enabled=true without any stonith resource configured? From my
understanding, the value of stonith-watchdog-timeout is only used during
the next fencing event and does not affect the currently running cluster.
Would it be possible to allow users some time to correct the configuration,
rather than causing the entire cluster to exit immediately?


Thanks,
xin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clusterlabs.org/pipermail/users/attachments/20250813/4df86777/attachment.htm>