David Wang created HDFS-10765:
---------------------------------

             Summary: Log a messages if finalize metadata hasn't been done, but 
customer kicked off HDFS Balancer
                 Key: HDFS-10765
                 URL: https://issues.apache.org/jira/browse/HDFS-10765
             Project: Hadoop HDFS
          Issue Type: Improvement
          Components: balancer & mover
            Reporter: David Wang


-- The finalize metadata hasn't been done yet, but there is no obvious message 
to inform that if started Balancer. Currently its just responding -7. Below is 
the logs.
--------------------------------------------------------------------------------------
2016-06-21 13:57:26,130 INFO [main] balancer.Balancer 
(Balancer.java:logUtilizationCollection(362)) - 1 over-utilized: 
[10.10.120.32:50010:DISK] 
2016-06-21 13:57:26,131 INFO [main] balancer.Balancer 
(Balancer.java:logUtilizationCollection(362)) - 5 underutilized: 
[10.10.120.93:50010:DISK, 10.10.120.97:50010:DISK, 10.10.120.95:50010:DISK, 
10.10.120.96:50010:DISK, 10.10.120.94:50010:DISK] 
2016-06-21 13:57:26,133 INFO [main] balancer.Balancer 
(Balancer.java:runOneIteration(526)) - Need to move 1.09 TB to make the cluster 
balanced. 
Jun 21, 2016 1:57:26 PM 0 0 B 1.09 TB -1 B 
Jun 21, 2016 1:57:26 PM Balancing took 1.519 seconds 

-- Copied the comments here.
// Should not run the balancer during an unfinalized upgrade, since moved
// blocks are not deleted on the source datanode.

http://opengrok.sjc.cloudera.com:8080/source/xref/CDH-5.5.1/hadoop/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/balancer/Balancer.java#530



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-dev-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-dev-h...@hadoop.apache.org

Reply via email to