Check data repair jobs
Before decommissioning a grid node, you must confirm that no data repair jobs are active. If any repairs have failed, you must restart them and allow them to complete before performing the decommission procedure.
If you need to decommission a disconnected Storage Node, you will also complete these steps after the decommission procedure completes to ensure the data repair job has completed successfully. You must ensure that any erasure-coded fragments that were on the removed node have been restored successfully.
These steps only apply to systems that have erasure-coded objects.
-
Log in to the primary Admin Node:
-
Enter the following command:
ssh admin@grid_node_IP
-
Enter the password listed in the
Passwords.txt
file. -
Enter the following command to switch to root:
su -
-
Enter the password listed in the
Passwords.txt
file.When you are logged in as root, the prompt changes from
$
to#
.
-
-
Check for running repairs:
repair-data show-ec-repair-status
-
If you have never run a data repair job, the output is
No job found
. You don't need to restart any repair jobs. -
If the data repair job was run previously or is running currently, the output lists information for the repair. Each repair has a unique repair ID.
Optionally, you can use the Grid Manager to monitor restoration processes in progress and display a restoration history. See Restore object data using Grid Manager. -
-
If the State for all repairs is
Completed
, you don't need to restart any repair jobs. -
If the State for any repair is
Stopped
, you must restart that repair.-
Obtain the repair ID for the failed repair from the output.
-
Run the
repair-data start-ec-node-repair
command.Use the
--repair-id
option to specify the Repair ID. For example, if you want to retry a repair with repair ID 949292, run this command:repair-data start-ec-node-repair --repair-id 949292
-
Continue to track the status of EC data repairs until the State for all repairs is
Completed
.
-