feat: add support for SummaryDataPoint in delta to cumulative processor

[lightme16]

Description

Adding support for Delta to Cumulatives for summary data points

Link to tracking issue

Fixes #38772

Testing

Added new unit tests for summary

Documentation

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: lightme16 / name: Alex (8b054cd, 3e7445f, b356e34, 4d3665b, 4ca46f0)

[atoulme]

Please add tests to cover the new functionality?

[lightme16]

Please add tests to cover the new functionality?

@atoulme, I can definitely do that. Thank you! However, before I invest further, could you please review the business logic? Are you satisfied with this functionality? Do you agree that sum & count metrics in the summary should be aggregated? I believe they should be, especially for our use case. However, I’m not sure if there are known issues that prevented this from being implemented in the past. If this functionality is indeed missing, we’re happy to add it, but I want to confirm with the community first.

More in the #38772

[lightme16]

@sh0rez, @ArthurSens, and @tedsuo: I just added a feature flag to disable the summary accumulation functionality by default. Please share your thoughts on the merge request. My company needs this to move forward, and it would be great to have it merged. Let me know if anything else broke while processing this for merging.

Thanks!

[github-actions]

This PR was marked stale due to lack of activity. It will be closed in 14 days.

[lightme16]

hello, any thoughts on MR - @sh0rez, @ArthurSens, and @tedsuo? thanks

[tombrk]

hey @lightme16, I apologize for the much too long waiting time. I was on pto and did not get back to this earlier.

while I do think you are onto a very valid use-case we should absolutely enable, I am not comfortable with deltatocumulative relying on behavior that is not part of the specification.

as per the spec, the Summary data point has no aggregation temporality. your example however shows it absolutely should have such information attached, because .sum and .count can clearly be delta.

ideally, I imagine the following:

• spec change to make SummaryDataPoint have AggregationTemporality
• make statsdreceiver set above to delta
• adapt and merge this PR

are you interested in continuing this work?

[lightme16]

Hello, @sh0rez I decided to move away from that summaries approach in my system due to this issue and use new native Prometheus histograms instead. I'm not interested in doing any more work on this, but I still believe it would be beneficial to have this patch merged and implemented. Perhaps someone else could take it on

[github-actions]

This PR was marked stale due to lack of activity. It will be closed in 14 days.

[github-actions]

Closed as inactive. Feel free to reopen if this PR is still being worked on.