[go: up one dir, main page]
Skip to content

Pull mirrors can fall into a "stuck" state shortly after production issues

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

Summary

It has been observed that pull mirrors can fall into a "stuck" state shortly after production issues on GitLab.com. The mirrors will stop updating and will not allow the users to force an update with the "update now" button simply stating "updating". This requires either support intervention in the form of a console escalation to force fail the import or the customer has to wait 24 hours for the pull to auto-fail. Both of these solutions are less than ideal, especially for customers relying on the mirror for production releases.

Steps to reproduce

This is hard to reproduce as it relies on production level issues on .com to be observed but there are plenty of example tickets to link.

Example Project

What is the current bug behavior?

The mirrors will stop updating and will not allow the users to force an update with the "update now" button simply stating "updating". This requires either support intervention in the form of a console escalation to force fail the import or the customer has to wait 24 hours for the pull to auto-fail.

What is the expected correct behavior?

The pull mirror will either fail or continue working. Ideally we would allow some method for a user to force fail an import for situations like this or to shorten the timeout window for the update.

Relevant logs and/or screenshots

example-issue

Output of checks

This bug happens on GitLab.com

Relevant Issues and Tickets (internal only)

Emergencies:

Tickets:

(There are more but these were some that stood out)

Issues:

(there are more these are the most recent)

Possible fixes

Add the ability to cancel the update action in the UI or shorten the timeout for failure on the update process.

Edited by 🤖 GitLab Bot 🤖