The answer to replace disks in Name Node is covered here - https://community.hortonworks.com/questions/3012/what-are-the-steps-an-operator-should-take-to-repl....
Just wondering if the same steps apply if we were to replace disks for other master services listed below :
- Secondary NameNode
- Journal Node
What is field recommendation around replacing disks?
Partner has implemented HA, but no RAID.
HA but no RAID
You are running multiple HBASE master. For HA, 3 ZK instance is must. RM HA is there. Multiple journal node is there. There is no secondary namenode in HA mode "Active or Standby namenode"
If cluster is configured in HA mode then you can replace disks with almost zero downtime.
Disclaimer: It needs to be well coordinated between various groups involeved doing the mainteance.