[SPARK-53405] Add metrics recording for latency of Spark app state transition #299
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…ansition ### What changes were proposed in this pull request? This PR adds metrics tracking state transition latency for Spark Application in format of ``` sparkapp.latency.from.<fromState>.to.<toState> ``` ### Why are the changes needed? Latency measuring would be useful to analyze the performance from scheduling / orchestration perspective. For example, to analyze which state causes significant overhad and therefore optimize at cluster/app level. ### Does this PR introduce _any_ user-facing change? More metrics becomes available. ### How was this patch tested? CIs. New unit test added. ### Was this patch authored or co-authored using generative AI tooling? No
|
cc @peter-toth for review |
|
Thanks @jiangzho for the PR. I am travelling for the rest of the week, but I can review this PR once I'm back. |
peter-toth
requested changes
Sep 1, 2025
| @@ -104,14 +98,17 @@ public void receivedEvent(Event event, Map<String, Object> metadata) { | |||
| getResourceClass(metadata); | |||
| final Optional<String> namespaceOptional = event.getRelatedCustomResourceID().getNamespace(); | |||
| resource.ifPresent( | |||
| aClass -> getCounter(aClass, action.name().toLowerCase(), RESOURCE, EVENT).inc()); | |||
| aClass -> | |||
| getCounter(getMetricName(aClass, action.name().toLowerCase(), RESOURCE, EVENT)) | |||
There was a problem hiding this comment.
As you extracted BaseOperatorSource and refactored getCounter() / getGauge() / getTimer() to have only one metricName argument, you need to call getMetricName() almost as many times as you call the above APIs.
I wonder if you could keep an old methods to avoid the changes at these callsites.
If you really need the new, one argument versions, then you can add them as overloaded methods.
|
The really like the idea, but let's try polish the implementation a bit. |
peter-toth
approved these changes
Sep 3, 2025
|
Thanks @jiangzho for the PR. Merged to |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What changes were proposed in this pull request?
This PR adds metrics tracking state transition latency for Spark Application in format of
Why are the changes needed?
Latency measuring would be useful to analyze the performance from scheduling / orchestration perspective. For example, to analyze which state causes significant overhad and therefore optimize at cluster/app level.
Does this PR introduce any user-facing change?
More metrics becomes available.
How was this patch tested?
CIs. New unit test added.
Was this patch authored or co-authored using generative AI tooling?
No