Very Uneven distribution of replicated data on a 2 unidirectional xdcr setup during initial loading phase.

Load 1M items on bucket1, cluster1
Load 1M items on bucket3, cluster2

Start replications from bucket1->bucket1 and bucket2-->bucket2
Unidirectional xdcr from cluster1 to cluster2 on bucket1
Unidirectional xdcr from cluster2 to cluster1 on bucket2

Always seeing 2 of the 4 nodes, getting replicated data faster on both the bucket/cluster combination.
The other 2 nodes get data after the replication on first 2 nodes is complete.

Can reproduce this on several tries.

Adding screenshots from the clusters.

Cluster1 : 10.3.121.31, bucket1 contains original 1M data
Cluster2: 10.3.121.38, bucket2 contains original 1M data

No other load on the system.

Load commands used are

Bucket1
nohup lib/perf_engines/mcsoda.py localhost:12233 vbuckets=1024 doc-gen=0 doc-cache=0 ratio-creates=1 ratio-sets=1 min-value-size=256 max-items=1000000 exit-after-creates=1 prefix=cb_one_1&

Bucket2
nohup lib/perf_engines/mcsoda.py localhost:22233 vbuckets=1024 doc-gen=0 doc-cache=0 ratio-creates=1 ratio-sets=1 min-value-size=256 max-items=1000000 exit-after-creates=1 prefix=cb_two_1&

We expect data to be uniformly replicated to all the nodes on the destination clusters?