[ https://issues.apache.org/jira/browse/HDFS-14894?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16946117#comment-16946117 ]
Kihwal Lee commented on HDFS-14894: ----------------------------------- We could sort nodes based on the utilization, so that highly utilized nodes get scheduled first. > Add balancer parameter to balance top N used nodes > -------------------------------------------------- > > Key: HDFS-14894 > URL: https://issues.apache.org/jira/browse/HDFS-14894 > Project: Hadoop HDFS > Issue Type: Improvement > Components: balancer & mover > Reporter: Leon Gao > Assignee: Leon Gao > Priority: Major > > We sometimes see a few of our datanodes reach very high usage (due to various > reasons) and we need to reduce their usage in an urgent situation. > We see two ways to achieve it currently, > -Calculate and reset balancing threshold. > -Pick nodes manually according to usage stats and put them in a file and use > `-resource` flag. > However, both of them are not very intuitive or too much manual work in an > urgent close-to-outage situation. Add a small feature to automatically pick > top N used hosts will be a straightforward option, for example `-top 10` to > only target top 10 used datanodes. -- This message was sent by Atlassian Jira (v8.3.4#803005) --------------------------------------------------------------------- To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org