When deploying a trek db migartion in k8s, sometimes two pods may run at the same time.. go-migrate seems to do nothing to avoid a race when doing a full db reset. We could consider letting trek obtain an advisory lock. Which is cluster-wide in postgres, not database-wide, so it will survive a database reset.
When deploying a trek db migartion in k8s, sometimes two pods may run at the same time.. go-migrate seems to do nothing to avoid a race when doing a full db reset. We could consider letting trek obtain an advisory lock. Which is cluster-wide in postgres, not database-wide, so it will survive a database reset.