[BUG] The Changelog Verifier GitHub Action runs 26 times and sends 26 emails if changelog is omitted

[dbwiddis]

Describe the bug

The Changelog Verifier GitHub Action triggers when a review is requested.

OpenSearch/.github/workflows/changelog_verifier.yml

Lines 1 to 4 in 2dc071f

	name: "Changelog Verifier"
	on:
	pull_request:
	types: [opened, edited, review_requested, synchronize, reopened, ready_for_review, labeled, unlabeled]

Because reviews are sent to every individual listed in CODEOWNERS, the action runs once for each maintainer. This spams the email of the contributor who neglected to include a change log.

One email is probably sufficient.

Related component

Build

To Reproduce

1. Submit a PR to OpenSearch that does not include a change log entry.
2. Check your email.
3. Count 26 emails.
4. Click delete 26 times, and submit a new commit with a change log.

Or you can review the GitHub action runs for every commit and see them scrolling past the 25-run pagination break:
https://github.com/opensearch-project/OpenSearch/actions/workflows/changelog_verifier.yml

Expected behavior

One email to a PR submitter who neglected to add a change log.

Even two may be tolerable.

Perhaps three if it's more important to ensure a changelog than prevent spamming contributors' inboxes.

26 is right out.

Additional Details

Additional context

It is likely that the review_requested trigger can be removed. I'm not sure how many of the other options are really required here.

[peternied]

[Triage - attendees 1 2 3 4 5 6 7 8]
@dbwiddis Thanks for filing, we'd look forward to a pull request to address this

[peternied]

Why 26?

When the pull request is created for each listed reviewer and event is fired, in fact I count 26 from this PR [1]. That might be the vast majority for that initial salvo.

• [1] Personalize code ownership areas #11600

I would recommend removing the following types that are subscribed to by this job:

• review_requested Large reduction in events off the bat
• ready_for_review Transition from draft->ready doesn't change the code, hence doesn't impact if the check will pass or fail
• reopened Same as ready for review.

[kotwanikunal]

@dbwiddis I have reduced the duplicate event triggers here: #13072