Retry initial indexing after a failure

Description

A repository stuck in the failed state never recovered. Nothing retried initial indexing after a failure.

This MR adds one automatic retry for initial code indexing:

  • A first failure moves the repository to a new pending_retry state.
  • A scheduled task retries pending_retry repositories every 30 minutes, with a full re-index.
  • If the retry also fails, the repository moves to failed and stays there.

Repositories that are already in the failed state predate this mechanism. We will take care of them in https://gitlab.com/gitlab-org/gitlab/-/work_items/629396.

Related to #548766 (closed).

Edited by Arturo Herrero

Merge request reports

Loading
Loading