[GOBBLIN-1648] Avoid Gobblin Cluster Manager downtime during DB failover

• Type: Task
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: gobblin-cluster
• Assignee: Hung Tran
• Reporter: Kip Kohn
• Created: 17/May/22 18:46
• Updated: 26/May/22 17:11

Description

When the DB fails over (so the currently connected instance which was active becomes a follower), the connection on hand becomes read-only.  Enforce verification when this happens, so we know to create new, replacement connections (to a read-write host).