Ensure Cluster is fit for production
This is about what we need to do for Gitaly Cluster to fit nicely into the gitlab.com production infrastructure.
For generic requirements and tooling that self-hosted customers would also benefit from, see gitlab-org&7896 (closed).
Scaling up
- Replication/reconciliation are currently global over all repos in a db
- Failure domains / production layout, virtual storages, databases etc
Monitoring and alerting, including SLOs
Debug workflows specific to how it's installed in production
(finding root-causes in Cluster with the tooling we have)
Training and documentation
Edited by Andras Horvath