Handling large backup restores on task runner
When restoring a backup to a Kubernetes installation, we use the
task-runner pod. For cases where the database and/or repositories are very large the
task-runner can run out of disk space.
A customer started out trying to restore a whole backup (all uploads, artifacts, etc) which was about 500GB. The uploads and similar items had already been migrated from the old Omnibus instance to object storage manually so I asked them to try taking a backup and to skip everything but the database and repositories. Still the 80GB file was too big.
How can we facilitate restoration of large backups in