A messages about host moving token may be observed in the
opscenterd.log file without any actual token movement operations having been initiated.
The following kind of message might be observed in the
2016-02-29 13:45:56+0100 [MyCluster] INFO: Host 10.1.2.3 moved from '4902817346077229087' to '5266222778472694838' 2016-02-29 13:45:56+0100 [MyCluster] INFO: Node 10.1.2.3 has multiple tokens (vnodes). Only one picked for display.
This is a bug caused by the fact that for nodes with vnodes enabled OpsCenter will pick a single token from the list as a representation of that node for internal purposes.
When OpsCenter checks for any potential changes to the cluster's topology it picks a token from the list in a non-deterministic way, causing it to think the node has moved.
This should not have any impact on OpsCenter functionality or the behavior of your cluster, and can be ignored.
The issue is outlined in the following internal OpsCenter Jira.
OPSC-7620 - Opscenter incorrectly detects a token has moved
At the time of writing this article there is no release date for this issue, but it will be addressed in a future release of OpsCenter.