what are the steps in order to replace faulty disk in datanode
we have ambari cluster - version 2.6.1
HDP - version 2.6.4
each datanode have 12 disks with 500G size
one of the disk is faulty disk and need to replace it
what are the full steps that required , in order to replace the faulty disk
Regarding your query: "the option Selected hosts isn't active ( in spite I see it )"
Here are the steps:
1. When you click on "Hosts" tab then a page opens with the list of hosts.
2. You will need to click on the "Checkbox" just in front of the hostnames mentioned in hosts page.
3. Then click on "Actions" --> "Selected Hosts" (you should see the number of hosts which you checked in step2)
4. Now you should be able to do:
Ambari UI -->Hosts(Tab)-->"Actions"(Drop down)--> Select the Host that you want --> then click on "Selected Hosts" --> DataNode --> Decommission
.
@Jay , now its clear, thanks to your great explain , when I do decommission , is it mean that datanode component can be up? , of before the decommision we need to stop anyway the componet on the datanode ?
@Jay , after we do the decommission on some datanode , do we need also to stop the components on that datanode ? and then replaced the disk , or maybe it is inoufgh to decomission without to stop the componet and then replaced the disk ?
(This is just partial answer to your last update)
Decommissioning DataNodes can be done in the following way for Ambari 2.6.1
Ambari UI --> Hosts (Tab) --> "Actions" (Drop down) --> All Hosts / Filtered Hosts (choose based on your need) --> Choose "DataNode" option --> Decommission
.
@Jay , but it will Decommission all datanode ? , and we want to Decommission only the datanode with the faulty disk ( am I right here ? ) , or some other option to Decommission only the specific datanode ( worker03 )
You can select the host which you want and has DataNode (Please use "Selected Hosts" option)
Ambari UI -->Hosts(Tab)-->"Actions"(Drop down)--> Select the Host that you want --> then click on "Selected Hosts" --> DataNode --> Decommission
.
This will ensure that the operations should be only performed to the list of host which you have selected using "
Below article help you to replacing faulty disks on datanode.
https://community.hortonworks.com/articles/3131/replacing-disk-on-datanode-hosts.html
Please accept the answer you found most useful.
any way why not just to the follwing steps:
1. stop all data node component
put the host in maintenance mode - I am not sure if this step is needed ?
2. shutdown the datanode machine
3. replace the fault disk
4. start the datanode machine
5. start the datanode components
@Michael Bronson
Decommissioning is a process that supports removing components and their hosts from the cluster. You must decommission a master or slave running on a host before removing it or its host from service. Decommissioning helps you to prevent potential loss of data or disruption of service. Below HDP documentation for Ambari-2.6.1 help you to decommission a DataNode. When DataNode decommissioning process is finished, the status display changes to Decommissioned.
I hope that the above answers your questions.
@Michael Bronson Correction: Replace "Filtered Host" with "Selected Hosts" option in my previous update.
