Hadoop/HDFS: Connection refused, all data nodes at 100%

Asked Jun 23 '16 at 21:49

Active Jun 23 '16 at 21:49

Viewed 162 times

Up until a few days ago I had a working Hadoop cluster. I copied over a very large data set and maxed out all of my datanodes. Running any sort of HDFS command gives me:

>hdfs dfs -ls
>ls: Call From hadoop-n2/XXX.XXX.XX.XX to hadoop-n2:8020 failed on connection exception: java.net.ConnectException: Connection refused; For more details see:  http://wiki.apache.org/hadoop/ConnectionRefused

So I can't go in a delete the files that are taking up space. I tried wiping HDFS clearn as described here but I don't see any tmp property in hdfs-site.xml. I tried reducing the replication factor from 3 to 1 but that didn't seem to change anything.

I'm running Hadoop v2.6.0 with Cloudera CDH v5.6.0.

edited May 23 '17 at 12:16

Community

asked Jun 23 '16 at 21:49

Sal

1,653
6
23
36

ssh keys might have changed – John Simon Jun 24 '16 at 07:38
Possible duplicate of : http://stackoverflow.com/questions/28661285/hadoop-cluster-setup-java-net-connectexception-connection-refused – Nishu Tayal Jun 24 '16 at 12:56
@JohnSimon Why would the ssh keys change? I'm able to ssh into any node from any other node. – Sal Jun 24 '16 at 15:31
Were you using Ambari? If so, I would have just added another datanode to the cluster to temporarily relieve the nodes. – Petro Jul 31 '18 at 18:10

Hadoop/HDFS: Connection refused, all data nodes at 100%

0 Answers0