I have recently setup HA on our cluster and noticed this warning. This appears to only happen on the standby namenode. Looking through some posts I was able to find that this occurs based on the order of the namenodes within the name service namespace (not sure if this is true). I manually failovered the namenodes several times and the warnings just rotated to the standby node. I also noticed within the hdfs-site.xml file that the dfs.ha.namenodes had added numbers to the names of the namenodes. Overall though the cluster functions correctly, I am able to utilize all tools/roles without errors.