🚚 The story behind helm-gitlab
helm-gitlab didn't start as a chart. It started as a migration: moving an entire self-hosted GitLab to another cloud — and then proving the move was lossless, byte for byte.
Pulling that off meant treating every backing service as something you own and can verify, not a black box: an in-cluster PostgreSQL HA pair you can checksum, SeaweedFS object storage you control, Valkey, and Traefik routing — all reproducible from source. This chart is the generalized, cleaned-up result of that work, so the next person doesn't have to rediscover the hard parts.
📖 Read the full write-up
The complete account — "The day I moved our entire GitLab to another cloud and proved I didn't lose a single byte" by Shane Froebel — is on Medium:
If you're evaluating whether to run your own GitLab this way, that story is the why behind every design decision documented here.