Github user zentol commented on the issue:
https://github.com/apache/flink/pull/5857
I I see the value of a `cluster_id` but you're mixing concerns here. It is
not the responsibility of a `reporter` to introduce a `cluster_id` tag.
Reporters are to faithfully report the metrics and their associated variables,
not add more. Instead we may want to think about adding a configurable
`cluster_id` value.
With that out of the way we have arrived at my previous question, random ID
vs actual ID. That the `job_name` will be equal to `tm/jm_id` is precisely why
i prefer this approach, it doesn't introduce additional noise in the tags.
I've already started working on `FLINK-9543` so we can proceed with this
PR; the issue must be extended to also expose an ID for Dispatchers and
register the JVM metrics for each JM. as to your question, `ResourceID`s are
exactly what we're looking for; this is also what we're using for
`TaskExecutor`.
---
