Document lifecycle of Gitaly pods running in Kubernetes during rotations/upgrades
Work to be done:
- Investigate and document how upgrades would work for Gitaly pods, how they respond during rotations or the self-healing process Kubernetes is known for doing.
- look into how rollbacks work : Performing a Rolling Update
- Determine which deployment strategy works best for Gitaly.
- For stateful applications which Gitaly is, normally have a persistent volume. If a blue-green strategy of deployment is used, migration will definitely cause some impact. Document the impact. Extension: documentating what state we have to store.