How to cleanup "Failed Actions" without restarting resources

[z_haseeb]

To trigger a failover I killed process id of application which is clustered/HA with pacemaker/pcsd. The application restarted again on same node successfully. I don't have any problem yet. But when I go to cleanup the "Failed Actions" , all resources restarts as I trigger pcs resource cleanup command. Please guide how can I cleanup Failed Actions which should not restart all resources.

[hunter86_bg]

What is your fencing mechanism ?
As per the following unverified solution such issue can be observed if your stonith device provide unfencing.
If it is the same for you - the issue is being investigated by Red Hat in Bugzilla #1427648 for RHEL 7 and #1427643 for RHEL 6.

[z_haseeb]

Fencing is disabled in our setup

[TrevorH]

You need fencing. You cannot run a cluster without it safely.

[z_haseeb]

You need fencing. You cannot run a cluster without it safely.

Without fencing we can run a cluster. Fencing is not a prerequisite for a cluster. However its a concert safety. Furthermore I am using cluster for an application(not a back end data) with more than 1 corosync heartbeats with bonded NICs so chances are less for split brain.
Discussion is going to another way. My question is very simple. I don't want to restart all resources if I want to clear the Failed Actions messages.

[TrevorH]

Fencing is not a prerequisite for a cluster

That is not what the pacemaker devs think and they are about to make fencing configuration mandatory. Fencing is not optional.

[z_haseeb]

Dear TrevorH. Thanks for your words but this does not seem to resolve my query in case I implement Fencing.

[hunter86_bg]

In order to simply reset the fail count , you can use:

Code: Select all

pcs resource failcount reset <resource id> [node]

Yet fencing is not only recommended but a kind of "mandatory". You can check the sbd fencing (aka "poison pill") , which requires only a shared device. If you don't have a shared device - you can use a CentOS iSCSI Target.

[z_haseeb]

Could not able to reset Failed Action thru below suggested commands

[quote="hunter86_bg"]In order to simply reset the fail count , you can use:

Code: Select all

pcs resource failcount reset <resource id> [node]

pcs resource failcount reset REDIS_monitor_45000 XXXCOM2
No failcounts needed resetting

pcs resource failcount reset REDIS XXXCOM2
No failcounts needed resetting

[hunter86_bg]

Could not able to reset Failed Action thru below suggested commands
No failcounts needed resetting

It seems the issue is not in the fail count.