hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jean-Daniel Cryans <jdcry...@apache.org>
Subject Re: HBase minimum block size for sequential access
Date Wed, 28 Jul 2010 15:36:22 GMT
Forced actions are currently async, and there's no feedback (this is
currently being worked on for 0.90). I personnally look at the region
servers' web UI metrics, when the compaction queue is at 0 on most
machines then you know it's done. Ganglia makes this very easy as you
can graph that for all the machines at the same time.

Compacting files involves rewriting them (into new files), so it is
expected that the DFS usage grows during that time. Also files on HDFS
are put in a "garbage can" and are only really deleted later.


On Wed, Jul 28, 2010 at 8:22 AM, Andrew Nguyen
<andrew-lists-hbase@ucsfcti.org> wrote:
> So, I ran the following command in the shells:
> alter 'tablename', {NAME=>'cfname', BLOCKSIZE=>1045876}
> major_compact 'tablename'
> How do I know the major compact completed successfully?  I saw that the number of regions
has grown quite a bit but I'm not quite sure to know when it's all finished so that I can
start the benchmarking.  Another thing that I noticed is after initiating the above, my DFS
usage has continued to grow.  I had 800g total, and 300g used.  After initiating the compaction,
I am now at 614M free...  Is this expected?
> Thanks!
> On Jul 27, 2010, at 10:13 AM, Jean-Daniel Cryans wrote:
>> After altering the table, issue a major compaction on it and
>> everything will be re-written with the new block size.

View raw message