Hi, after some updates I want to reboot our hbase (2.1.4)/hadoop(3.1.2) cluster. Its 8 nodes, 3 master server (2 backup masters) and 5 region servers Hadoop is a HA setup, datanodes and region servers on same machines.
What is the preferred way of a rolling restart? I have read the docu about it, but I'm unsure. Is this a correct way? * first the region servers, for each region server: -- stop the balancer -- graceful stop the region server -- stop the datanode (hadoop-daemon stop ...) -- reboot node -- start datanode (hadoop-daemon start ...) -- start region server (hbase daemon start...) *then for the master server (and namenodes) - stop backup master - stop namenode (on node where the backup master was stopped) - reboot node - start namenode - start master server -do this for the second backup master again - do this for the active master On the namenodes the are the journal nodes running, at what point do I have to stop / start them? Or are they started by the namenode itself? Thanks for helping Michael