This kind of debugging is outside the scope of the discussion forum capabilities. You should reach out to Aerospike directly and see how best this can be resolved. Or you can email me at pgupta@aerospike.com and I can get the right people to talk to you.
That’s an interesting interpretation, but it is more complicated. Some of these are as you described. There can also be objects that fit your description that are ‘master’ or ‘prole’. In the case that you add more than RF nodes, only half of the ‘non_replica_objects’ are to be transmitted, the other half is to maintain your replication-factor while migrating a partition.
I would recommend monitoring your ‘migrate_partitions_remaining’ stat instead. You could derive a approximate ceiling for the number of objects to be transmitted per node with the formula: ((objects * cluster_size) / 8192.0) * (partitions_remaining_tx + partitions_remaining_rx).
Ok, I’ll contact if necessary.
Thanks, your formula is more accurate, I’ll consider that.