Details
Description
Scenario
--------------
1. Create two clusters with 2 nodes each and create bi-xdcr on 2 buckets. Load data, watch replication. Pause all replications at C1. C2 continues to replicate to C1.
2. Rebalance-in one node at cluster1 while failing over one node and rebalancing it out at cluster2. Resume all replications at C1.
3. Notice that on cluster2, all ongoing replications go from "replicating" to "starting up" mode and there's no outbound replication category for any of the cluster buckets.
Setup
--------
[Cluster1]
172.23.106.45
172.23.106.46 <--- 172.23.106.209 [rebalance-in]
[Cluster2]
172.23.106.47
172.23.106.48 ---> failover and rebalance-out
Reproducible?
---------------------
Yes, consistently, tried thrice.
Attached
--------------
cbcollect info and screenshot
Script
--------
./testrunner -i /tmp/bixdcr.ini -t xdcr.pauseResumeXDCR.PauseResumeTest.replication_with_pause_and_resume,items=30000,rdirection=bidirection,ctopology=chain,sasl_buckets=1,rebalance_in=source,rebalance_out=destination,failover=destination,pause=source
Will scale down to one replication and also try with xmem.