Description
- 3 node cluster (using v. lightweight vagrants)
- Set autofailover to 5 seconds
- Killed one of the VMs (node3 - 10.142.170.103)
- Node was successfully failed over.
- Reinstantiated the node3 VM
- Any attempts to re-add the node to the cluster were denied, citing "Failed to reach erlang port mapper". Tried restarting the couchbase service on node3 and same response.
- After adding a new node in and rebalancing, I was then able to re-add node3 successfully
My suspicion is that the re-add of node3 was denied until it had cleanly removed it from the cluster (new VM with same IP causing confusion?). However it would be more desirable to allow the node to be re-added without having to invoke a rebalance.
Logs:
https://s3.amazonaws.com/cb-engineering/davidH/collectinfo-2017-09-06T205740-ns_1%4010.142.170.101.zip
https://s3.amazonaws.com/cb-engineering/davidH/collectinfo-2017-09-06T205740-ns_1%4010.142.170.102.zip
https://s3.amazonaws.com/cb-engineering/davidH/collectinfo-2017-09-06T205740-ns_1%4010.142.170.103.zip
https://s3.amazonaws.com/cb-engineering/davidH/collectinfo-2017-09-06T205740-ns_1%4010.142.170.104.zip