Implement remediation steps from recent Zoekt incident
Summary
We've had a Zoekt incident gitlab-com/gl-infra/production#18213 (comment 1974308260), and I think we might have a bug with the node_assignment
task.
We've been trying to assign the same namespace over and over again.
My initial guess is that this line causes the problem https://gitlab.com/gitlab-org/gitlab/-/blob/3293eef830b3d3f145b6f13fa8b373b9ebe072e1/ee/app/services/search/zoekt/scheduling_service.rb#L18
update 2024-08-08: We believe this happened due to lack of support for large namespaces. This MR should help fix this issue: !161212 (merged)
Steps to reproduce
What is the current bug behavior?
What is the expected correct behavior?
Relevant logs and/or screenshots
Possible fixes
Edited by John Mason