Re: Hive-Hbase with large number of columns

2010-06-17 Thread Ray Duong
IONS.PART_NAME, which is an indexed column. Try again, undoing that >> line of the patch first. >> >> JVS >> >> From: Ray Duong [ray.du...@gmail.com] >> Sent: Wednesday, June 16, 2010 5:24 PM >> To: hive-user@hadoop.

Re: Hive-Hbase with large number of columns

2010-06-16 Thread Carl Steinbach
again, undoing that > line of the patch first. > > JVS > > From: Ray Duong [ray.du...@gmail.com] > Sent: Wednesday, June 16, 2010 5:24 PM > To: hive-user@hadoop.apache.org > Subject: Re: Hive-Hbase with large number of columns > > H

RE: Hive-Hbase with large number of columns

2010-06-16 Thread John Sichi
@hadoop.apache.org Subject: Re: Hive-Hbase with large number of columns Hi, I applied the patch HIVE-1364 and rebuilt the metastore. I was able to create an external table in Hive for a large number of columns ( upto 4000bytes). Now when I tried to drop the external table I get the following

Re: Hive-Hbase with large number of columns

2010-06-16 Thread Ray Duong
Hi, I applied the patch HIVE-1364 and rebuilt the metastore. I was able to create an external table in Hive for a large number of columns ( upto 4000bytes). Now when I tried to drop the external table I get the following error message. Is there another file that I need to modify in order to dro

Re: Hive-Hbase with large number of columns

2010-06-15 Thread John Sichi
Column-level properties are attractive for other reasons, but I don't think we should do it as a workaround for underlying limits. I've noted in JIRA that I think a LOB would be more appropriate here. Note that while you're waiting for a permanent resolution, you can use ALTER TABLE on your me

Re: Hive-Hbase with large number of columns

2010-06-15 Thread Ray Duong
Yes, I think I might have to do that. I was trying to avoid multiple Hbase scans with one big table. BTW, would it better to store the column SERDE properties for each column versus at the table level to avoid the 767 or 4000 byte limitation? Thanks again, -ray On Tue, Jun 15, 2010 at 2:42 PM,

Re: Hive-Hbase with large number of columns

2010-06-15 Thread Edward Capriolo
On Tue, Jun 15, 2010 at 5:04 PM, Ray Duong wrote: > Thanks for all the help. > > -ray > > > On Tue, Jun 15, 2010 at 1:26 PM, Carl Steinbach wrote: > >> Hi Ray, >> >> 4000 bytes is the maximum VARCHAR size allowed on Oracle 9i/10g/11g. As >> far as I know this is the smallest maximum VARCHAR size

Re: Hive-Hbase with large number of columns

2010-06-15 Thread Ray Duong
Thanks for all the help. -ray On Tue, Jun 15, 2010 at 1:26 PM, Carl Steinbach wrote: > Hi Ray, > > 4000 bytes is the maximum VARCHAR size allowed on Oracle 9i/10g/11g. As far > as I know this is the smallest maximum VARCHAR size out of the databases we > currently try to support (MySQL, Oracle,

Re: Hive-Hbase with large number of columns

2010-06-15 Thread Carl Steinbach
Hi Ray, 4000 bytes is the maximum VARCHAR size allowed on Oracle 9i/10g/11g. As far as I know this is the smallest maximum VARCHAR size out of the databases we currently try to support (MySQL, Oracle, Derby, etc). Carl On Tue, Jun 15, 2010 at 1:15 PM, Ray Duong wrote: > Thank John/Carl, > > Ye

Re: Hive-Hbase with large number of columns

2010-06-15 Thread Ray Duong
Thank John/Carl, Yep, there seems to be a limit on the 767 byte size. So I see the patch HIVE-1364 to set it to 4000 bytes. I'm using Db-derby, do you know if there is a limit beyond 4000 bytes? -ray Error: Caused by: ERROR 22001: A truncation error was encountered trying to shrink VARCHAR 'se

Re: Hive-Hbase with large number of columns

2010-06-15 Thread Carl Steinbach
Hi Ray, There is currently a 767 byte size limit on SERDEPROPERTIES values (see http://issues.apache.org/jira/browse/HIVE-1364). It's possible that you're bumping into this limitation (assuming you abbreviated the column names in your example). On Tue, Jun 15, 2010 at 12:03 PM, John Sichi wrote:

Re: Hive-Hbase with large number of columns

2010-06-15 Thread John Sichi
That exception is coming from the metastore (trying to write the table definition). Could you dig down into the Hive logs to see if you can get the underlying cause? You can get the logs to spew on console by adding "-hiveconf hive.root.logger=DEBUG,console" to your Hive CLI invocation. JVS

Hive-Hbase with large number of columns

2010-06-15 Thread Ray Duong
Hi, I'm trying to map a Hbase table in Hive that contains large number of columns. Since Hbase is designed to be a wide table, does Hive/Hbase integration have any set limitation on the number of columns it can map in one table? I seem to hit a limit at 10 columns. Thanks, -ray create external