hadoop-mapreduce-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Deepak Nagaraj <n.dee...@gmail.com>
Subject du says 3X more used than dfsadmin
Date Mon, 03 Oct 2016 19:58:03 GMT

I'm running a single-node test HDFS cluster (RF=1, in case that matters).

Here's my problem:

* hdfs -dfsadmin shows 10G for "dfsUsed".
* But "du -sh" on the data directory (/var/hadoop_data/hdfs/datanode) shows 30G.

I also noticed that the following file has around ~30G for dfsUsed:

$ cat dfsUsed
32516294656 1474677109034
$ pwd

So I'm wondering if dfsadmin is not picking it up correctly?

The end result is that a script that's supposed to archive old data is
not doing anything, because it thinks there's more space, whereas in
reality there isn't any.

Can someone please help me here?

Thanks in advance,

To unsubscribe, e-mail: user-unsubscribe@hadoop.apache.org
For additional commands, e-mail: user-help@hadoop.apache.org

View raw message