On clustered engines

[bohledevs]

Hello there Stephen,

Thank you for this amazing contribution.

I was wondering if your plugin would be applicable to a clustered engine, since I had some consistency issues trying to apply it to multiple Camunda Spring-Boot nodes.

I know it works perfectly on Camunda Run. Just trying to find a workaround to take it to a clustered architecture.

Best Regards,
Facundo Bohle

[StephenOTT]

You only need to run the plugin on 1 node. Think of the Monitoring plugin as a dedicated node (at scale / in larger cluster).
This is why the sample code has the Job Executor disabled: https://github.com/StephenOTT/Camunda-Monitoring#camunda-run-quick-start-with-prometheus. It is a node which never does anything for the overall engine except for running monitoring queries.

[bohledevs]

This really clarified my understanding of the plugin.
I was attaching them to every node in the cluster and it was producing incorrect data. This was a mistaken notion I've been carrying since your previous plugin.

Thank you again!

[StephenOTT]

The one area where you may want to run the plug-in on every node is if you are counting occurrences on the specific node: such as intercepting a Command, counting durations of executing scripts or other runtime specific actions: you would count these in memory as they occur and then scrape the data across all nodes