#opendaylight-clustering: “clustering hackers"

Meeting summary

1. 1. Mark presenting slides on 2-Node HA (moizer, 15:14:35)
   2. 2-Node scenario is for mostly HA only (moizer, 15:18:25)
   3. colin says we design the solution based on the assumption that we cannot have a partition (moizer, 15:25:28)
   4. colin says the raft implementation should be changed to work with a 1-Node cluster if leadership could be decided on a secondary heartbeat (moizer, 15:27:06)
   5. Robert asks if this needs to be a openflow specific solution (moizer, 15:28:16)
   6. Mohnish is of the view that the backup should not automatically take over as leader (moizer, 15:32:27)
   7. a 2-way heartbeat may be needed (primary and backup may need to send heart beats to each other) (moizer, 15:42:23)
   8. Mark says we should assume that openflow switches are connected to both the primary and secondary nodes (moizer, 15:50:07)
   9. we need to assume that the openflow switches must be 1.3 (moizer, 15:51:07)
   10. Moiz wants to see a design which shows how do we alter the Raft based design to accomodate the 2-Node HA configuration (moizer, 15:58:07)
   11. ACTION: Mark to post slides to the wiki (moizer, 16:03:40)
   12. ACTION: Mark to start conversation on mailing list on 2-Node HA design (moizer, 16:10:12)

Action items

1. Mark to post slides to the wiki
2. Mark to start conversation on mailing list on 2-Node HA design

People present (lines said)

1. moizer (14)
2. odl_meetbot (3)
3. tbachman (1)