Hi @clvi, Try adding the --appOwner <username> to the yarn logs command. However,  I think the application states are erased from the RM state store, probably due to an RM State Restore. ... View more

Thx @bsani We have sat HA on our name nodes while we don't want the cluster to be unawailable. So is there a best practice for doing patching on a cluster that is supposed to be available 24/7. How to avoid rebalancing during patching When upgrading datanodes in chuncks, is there a way to make sure that replica of a data block is available on one of the servers alive /Claus ... View more