[yordis]

• Postgrex drops a notification that arrives while the notifications connection is running its idle ping, and that is only fixed on Postgrex main (Fix notifications lost during idle ping elixir-ecto/postgrex#795), so no released version protects us
• A 100ms ping cadence turned that window into a lost append roughly once every 15 runs of the subscription tests, which is why CI failed at random with an empty mailbox and no log output to point at
• No other test environment sets an idle interval, and neither of the two reconnect tests that motivated it needs one

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Advanced

Run ID: 55a21ae9-73ee-4a42-abf1-e3d8c4ceaa14

📥 Commits

Reviewing files that changed from the base of the PR and between 9d360e3 and 6f00234.

📒 Files selected for processing (1)

• config/test.exs

💤 Files with no reviewable changes (1)

• config/test.exs

Included review availability: Your plan provides up to 1 included review per hour; 0 remain after this review.

---

Walkthrough

The test event-store configuration removes the explicit idle_interval value. The configured stores now use the serializer/Connection default idle interval.

Changes

Test event-store configuration

Layer / File(s)	Summary
Connection idle interval config/test.exs	Removed idle_interval: 100 from default_config. Other event-store settings remain unchanged.

Priority: ⬇️ Low

Estimated code review effort: 1 (Trivial) | ~2 minutes

Change: Bug fix

Merge Risk: ⚪ Minimal · up to 6f002

The test configuration now avoids the frequent idle ping that caused intermittent notification delivery failures. No actionable merge risk remains.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly identifies the test configuration change and its purpose: preventing notification loss during the idle ping. It is concise and specific.
Description check	✅ Passed	The description directly explains the Postgrex issue, the intermittent test failures, and why removing the 100 ms idle interval resolves the problem.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check. Docstring coverage is scoped to functions touched by this diff. Analyzed 0 functions across 0…
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch yordis/test-notifications-idle-ping

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{A rabbit checks the store at night
The idle clock now rests just right
No hundred ticks remain
Defaults guide the train
Config hops clean and light}

_{Comment @coderabbitai help to get the list of available commands.}

[cursor]

PR Summary

Low Risk
Test-only configuration change to reduce CI flakiness; no production or runtime behavior change.

Overview
Removes idle_interval: 100 from the shared TestEventStore test config in config/test.exs, so the test pool no longer forces a 100ms Postgrex idle ping.

That ping overlapped with a Postgrex bug where notifications arriving during the idle check can be dropped (fix only on unreleased main), which caused intermittent subscription test failures—empty mailboxes with no useful logs—about once every ~15 runs in CI. Other test stores and reconnect tests did not need this setting.

^{Reviewed by Cursor Bugbot for commit 6f00234. Bugbot is set up for automated code reviews on this repo. Configure here.}