[go: up one dir, main page]

Skip to content

CommitDiff & DiffStats timeout for particular commit on GitLab.com

Everyone can contribute. Help move this issue forward while earning points, leveling up and collecting rewards.

Summary

A commit with dozens/hundreds of long-named .zip files, plus many text files changes with thousands of +/- lines (whole Python project with test fixtures) is causing Gitaly to run into its medium/30sec timeout.

This bug report is a copy of the internal section-dev-request-for-help#149.

In MR, commit was not the HEAD of the that branch and pipeline didn't trigger.

The repository check reports no problem and git-sizer locally also doesn't report the commit as problematic.

Steps to reproduce

  1. Check mentioned commit 00e… in this internal customer ticket and load that page as an admin user > 500 error
  2. Check same commit within context of its MR. > Commit shows up in commits list, and /diffs?commit_id=00e page, but not in the MR's activity.

Example Project ☝️

What is the current bug behavior?

500 error in Sentry (no equivalent found in new-sentry 🤔) & ES for stacktrace with lib/gitlab/gitaly_client/diff_stitcher.rb, and file already closed/parse failure: read diff header line errors.

New sentry: https://new-sentry.gitlab.net/organizations/gitlab/issues/?environment=gprd&project=3&query=is%3Aunresolved+%22Projects%3A%3ACommitController%23show%22+%224%3ADeadline+Exceeded.%22&referrer=issue-list&statsPeriod=3d (by @vyaklushin)

Related: 500 error when loading older in-process Merge R... (#28424 - closed) (that one is about conflict_files_stitcher.rb & 55sec timeout, though) & Reusable Rapid Diffs (was: New diffs) (&11559)

What is the expected correct behavior?

  • Better error message, see #27817

Relevant logs and/or screenshots

Output of checks

This bug happens on GitLab.com

Possible fixes

Edited by 🤖 GitLab Bot 🤖