Fixed incoming recommendations disappearing on transient errors

[kevinansfield]

Summary

During boot Ghost attempts to fetch recommendations to validate that they still exist. If the site isn't reachable then the recommendation is soft-deleted causing it to disappear from various UIs with no way for the site owner to recover it themselves. On next boot Ghost will revalidate the recommendations and restore any that are back online but it does leave periods where transient network issues can confusingly make recommendations disappear.

Changes

• Transient fetch errors (429, 503) during incoming recommendation revalidation are now ignored instead of immediately deleting the mention
• Hard failures now require 3 consecutive occurrences before a mention is soft-deleted, tracked via a new revalidation_failure_count column on the mentions table
• A successful metadata fetch resets the failure counter to zero
• All failures are logged with a [Webmention Revalidation] prefix for easy diagnosis

ref https://linear.app/ghost/issue/ONC-1511/

[coderabbitai]

Important

Review skipped

Draft detected.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch worktree-ONC-1500-missing-recommendation-notifications

---

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

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

[ErisDS]

🤖 Velo CI Failure Analysis

Classification: 🔴 HARD FAIL

• Workflow: CI
• Failed Step: Determine changed packages
• Run: View failed run
  What failed: GitHub API rate limit exceeded, causing failure to list changed files
  Why: The error message indicates that the GitHub API rate limit has been exceeded, which is an infrastructure issue outside the control of the code. This is a hard failure that requires investigating the GitHub API usage and potentially increasing the rate limit.
  Action:
  Check the GitHub API usage and rate limit, and consider increasing the rate limit or implementing caching to avoid hitting the limit during CI runs.

[ErisDS]

🤖 Velo CI Failure Analysis

Classification: 🔴 HARD FAIL

• Workflow: Migration Review
• Failed Step: Run actions/github-script@v7
• Run: View failed run
  What failed: Unexpected end of JSON input from GitHub API
  Why: The error message indicates that the GitHub API returned an unexpected response, which is likely an infrastructure issue on GitHub's side. This is not a code issue, as the error is not related to the application code itself.
  Action:
  This is likely a temporary issue with the GitHub API. Try rerunning the workflow, and if the issue persists, you may need to investigate further on the GitHub infrastructure side.

[ErisDS]

🤖 Velo CI Failure Analysis

Classification: 🟠 SOFT FAIL

• Workflow: CI
• Failed Step: Run unit tests
• Run: View failed run
  What failed: Database schema has changed, but migration test failed
  Why: The root cause is a test failure related to the database schema integrity. This is a code issue, as the test is checking that the database schema has not changed without a corresponding migration being created.
  Action:
  The author should review the database schema changes, ensure a proper migration has been created, and update the test to reflect the new expected schema hash.