Backing up GitLab secrets and configuration
Why gitlab-secrets.json and gitlab.rb need their own backup, what breaks when the secrets file is lost after a restore, and keeping SSH host keys and TLS material for full rebuilds.
GitLab backup and disaster recovery
Specific backup, restore and Geo procedures for self-managed GitLab beyond the basic backup task: secrets, large instances, object storage and failover.
GitLab incremental repository backups
Cutting backup windows with incremental repository backups: how increments chain to a previous backup ID, what stays full, and why deleting increments breaks restores.
GitLab server-side repository backups to object storage
Letting Gitaly stream repository backups directly to object storage instead of the backup archive, the configuration it needs and how it lowers network load on large instances.
Backup strategy for large GitLab instances
What to do when the database passes 100 GB or Git data passes 400 GB: snapshots with read-only mode, Geo, file system transfers and splitting repository backups across storages.
Backing up GitLab data in object storage
Covering artifacts, LFS, uploads and registry blobs that the GitLab backup command skips when they live in S3, GCS or Azure, using provider versioning and replication instead.
Restoring GitLab backups: version and edition rules
Planning restores around the same-version, same-edition rule: keeping matching packages, restoring secrets first, and testing restores before an incident or migration.
GitLab Geo planned failover
Running a planned switchover to a Geo secondary: maintenance mode, letting replication catch up, promotion commands for single and multi-node sites and re-adding the old primary.
GitLab Geo for distributed teams and CI offload
Using Geo secondaries as local read caches for remote offices and runners: what is served locally, what is proxied to the primary and the network ports to open between sites.
GitLab Geo with a unified URL
Serving all Geo sites behind one GitLab hostname with location-aware DNS such as Route 53 or Cloud DNS, and the certificate handling the shared domain requires.
GitLab disaster recovery options compared: backups, snapshots and Geo
Choosing a recovery approach by instance size and tolerance for data loss: backup archives, disk snapshots, Geo warm standby and how they combine, since Geo still needs backups.
