[ClusterLabs Developers] Need advice on troubleshooting Custom Resource Agent

Jehan-Guillaume de Rorthais jgdr at dalibo.com
Tue May 5 16:49:24 UTC 2020


On Sun, 3 May 2020 15:42:50 +0000
Taylor <tschneider at live.com> wrote:
> Thanks for the reply.

You are welcome

> How do you force the promotion?
> 
>> I log into a node and run "crm_resource --resource named --force-promote"

ok

> Do you raise messages using ocf_exit_reason when your RA returns errors?
> 
>> No, I did not know about that functionality but I have added that to my
>> code.

great

> Could you put Pacemaker in debug to gather more logs
> 
>> Done

great, it should give much more informations

> Could you provide a small part of log showing the messages when you try to
> promote your resource?
> 
>> Sure. Attached are the logs when the resource is added. Attached are the
>> logs when I try to promote.

Sorry, I forgot to mention we need logs from the failing node AND the DC, where
cluster decisions are taken. Based on your pcs status, this was os004dns003:
"Current DC: os004dns003"

> Could you share your RAs code?
> 
>  - Sure. The code is available here: https://github.com/bdonnahue/pfra/ . In
> a nutshell, the RA is parameterized with script paths. It will execute a
> specified script when it is asked to perform a Resource Agent
> Operation/Action. For example, the monitor.sh script is executed when the
> Monitor Operation is executed. I am testing the RA with the BIND daemon and
> the scripts I pass to the pfra RA are in the scripts directory.

It looks like an AI on top of an API. What was your need?

There is an existing named resource agent, maybe you could extend it to suit
your need or even contribute to it?

Regards,



More information about the Developers mailing list