accumulo-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Billie Rinaldi <bil...@apache.org>
Subject Re: How Do I Pragmatically Know When A Compaction Is Over? (i.e., how do I find hotspots)
Date Wed, 02 Jan 2013 15:58:58 GMT
On Wed, Jan 2, 2013 at 7:12 AM, David Medinets <david.medinets@gmail.com>wrote:

> I have a map-reduce job which uses AccumuloInputFormat. Some of the
> mappers take 5 minutes while others take 40 minutes. Looking at the
> entry count it seems like some of the tablets have more entries than
> others. I'd like to generate a histogram of the number of entries per
> tablet.
>
> On the way to that goal, I learned that using Bulk Ingest does not
> update the Number of Entries so that I need to perform a compaction
> before I can learn the number of entries... Which leads me to how can
> I tell that a compaction is complete?
>

If you use the "-w" flag to the compact command, the command won't return
until the compaction has completed.  There's also a boolean wait parameter
for one of the TableOperations.compact methods.  Otherwise, you have to
look at the monitor page to see when tablets are no longer being compacted.

Billie

Mime
View raw message