[jira] [Created] (HDDS-15031) [Website v2][Troubleshooting]Slow datanode decommission

["Wei-Chiu Chuang (Jira)" <ji...@apache.org>]

Wei-Chiu Chuang created HDDS-15031:
--------------------------------------

             Summary: [Website v2][Troubleshooting]Slow datanode decommission
                 Key: HDDS-15031
                 URL: https://issues.apache.org/jira/browse/HDDS-15031
             Project: Apache Ozone
          Issue Type: Sub-task
            Reporter: Wei-Chiu Chuang


In certain scenarios, a datanode can take a long time to be ready for decommission.

A datanode transitions from DECOMMISSIONING to DECOMMISSIONED state if all containers are sufficiently replicated. For a dense datanode, that means hours to even days depending on the cluster size and network speed.

Execute command 'ozone admin datanode status decommission' to monitor the status of decommission. Use Grafana to check out decommission and replication related metrics.

If it feels slow, a few configuration properties can be bumped up to speed up decommissioning. hdds.scm.replication.datanode.replication.limit in SCM limit for concurrent replication commands that the SCM will send to a single Datanode;  hdds.datanode.replication.outofservice.limit.factor sets the weight of decommissioning commands over other commands from SCM.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@ozone.apache.org
For additional commands, e-mail: issues-help@ozone.apache.org