## Background Indexing of a repo could fail for any reason; one reason could be a huge number of files to index. But with time, the number of files could be reduced. So, we should have a mechanism to retry indexing the `failed` zoekt_repository. ## Proposal Add a `zoekt:rake_task` to move all failed zoekt_repository to `pending` with `retries_left` set to `1`. This rake task could accept the optional argument, an array of project IDs. If the project_ids argument is supplied, then it should process only those `failed` ids; otherwise, process all `failed` repos. Also, add documentation for this change. <!-- Please add a label for the type of maintenance as per https://handbook.gitlab.com/handbook/product/groups/product-analysis/engineering/metrics/#work-type-classification -->