Thanks Andy
On Tue, Jun 28, 2011 at 10:00 AM, Andrew Purtell wrote:
>> From: "Gan, Xiyun"
>
>> I still have another question, how to remove the partitions_$timpstamp
>> files produced in the HFileOutputFormat?
>
>
> One option is to accept a configuration option for the partitions file:
>
> -
> From: "Gan, Xiyun"
> I still have another question, how to remove the partitions_$timpstamp
> files produced in the HFileOutputFormat?
One option is to accept a configuration option for the partitions file:
- Path partitionsPath = new Path(job.getWorkingDirectory(),
- "partitions_"
Thanks Andy and Stack.
I'm not a native speaker, sorry for my ambiguous question. Andy
interpreted my question rightly.
I still have another question, how to remove the partitions_$timpstamp
files produced in the HFileOutputFormat?
On Tue, Jun 28, 2011 at 6:26 AM, Stack wrote:
> On Mon, Jun 27,
On Mon, Jun 27, 2011 at 3:09 PM, Andrew Purtell wrote:
> I think the sender asked if HBASE-1861 is the right approach.
>
Thanks Andy. Sorry for answering question w/ a question (its my genes).
We committed HBASE-1861 so in my thinking, HBASE-1861 is one
"legitimate" approach. I was thinking if
I think the sender asked if HBASE-1861 is the right approach.
- Andy
- Original Message -
> From: Stack
> To: user@hbase.apache.org
> Cc:
> Sent: Monday, June 27, 2011 2:51 PM
> Subject: Re: Multi-family support for Bulkload
>
> On Sun, Jun 26, 2011 at 8:42
On Sun, Jun 26, 2011 at 8:42 PM, Gan, Xiyun wrote:
> Take no account of HBASE-1861, is that a right way to use one
> RecordWriter for each column family to write separate hfiles?
>
What would you suggest instead?
St.Ack
HFileOutputFormat supports only one family in official distribution,
and HBASE-1861 added multi-family support.
Take no account of HBASE-1861, is that a right way to use one
RecordWriter for each column family to write separate hfiles?
Thanks
--
Best wishes
Gan, Xiyun