Hi, all

    My cluster uses hbase 0.92.0.
    I'm now trying to split some big regions manually. (I configured
region file size to 20G)
    I ran it like this:
hbase org.apache.hadoop.hbase.util.RegionSplitter statistic_visit1 -r -o 2

    Before I split, the table statistic_visit1 has only one region:
[hadoop@hadoop01 ~]$ hadoop dfs -du /hbase/statistic_visit1
Found 3 items
5161         hdfs://hadoop01:51001/hbase/statistic_visit1/.tableinfo.0000000001
0            hdfs://hadoop01:51001/hbase/statistic_visit1/.tmp
58234791267  hdfs://hadoop01:51001/hbase/statistic_visit1/e15d089574adaeacaf75fd8e1277ce2d

    And after the split is done,
[hadoop@hadoop01 ~]$ hadoop dfs -du /hbase/statistic_visit1
Found 6 items
5161         hdfs://hadoop01:51001/hbase/statistic_visit1/.tableinfo.0000000001
0            hdfs://hadoop01:51001/hbase/statistic_visit1/.tmp
58234783527  hdfs://hadoop01:51001/hbase/statistic_visit1/0eae290663e59659fc728333d0b18b47
287          hdfs://hadoop01:51001/hbase/statistic_visit1/7865a928112751bcf27fce2cc8304441
80           hdfs://hadoop01:51001/hbase/statistic_visit1/_balancedSplit
58234787479  hdfs://hadoop01:51001/hbase/statistic_visit1/e15d089574adaeacaf75fd8e1277ce2d

    Does the split result look fine? And how can I get more regions?

    And is there any danger, if I resize the region file size setting
to a smaller value? (like 5G)

    Thanks.
--

Regards,
Cheng Su

Are the options 'firstrow' and 'lastrow'  what you want?

I have noticed that the '_balancedSplit' file was still there, so ...
you must got this:
https://issues.apache.org/jira/browse/HBASE-7138

On Fri, Nov 9, 2012 at 3:47 PM, Cheng Su <[EMAIL PROTECTED]> wrote:
> Hi, all
>
>     My cluster uses hbase 0.92.0.
>     I'm now trying to split some big regions manually. (I configured
> region file size to 20G)
>     I ran it like this:
> hbase org.apache.hadoop.hbase.util.RegionSplitter statistic_visit1 -r -o 2
>
>     Before I split, the table statistic_visit1 has only one region:
> [hadoop@hadoop01 ~]$ hadoop dfs -du /hbase/statistic_visit1
> Found 3 items
> 5161         hdfs://hadoop01:51001/hbase/statistic_visit1/.tableinfo.0000000001
> 0            hdfs://hadoop01:51001/hbase/statistic_visit1/.tmp
> 58234791267  hdfs://hadoop01:51001/hbase/statistic_visit1/e15d089574adaeacaf75fd8e1277ce2d
>
>     And after the split is done,
> [hadoop@hadoop01 ~]$ hadoop dfs -du /hbase/statistic_visit1
> Found 6 items
> 5161         hdfs://hadoop01:51001/hbase/statistic_visit1/.tableinfo.0000000001
> 0            hdfs://hadoop01:51001/hbase/statistic_visit1/.tmp
> 58234783527  hdfs://hadoop01:51001/hbase/statistic_visit1/0eae290663e59659fc728333d0b18b47
> 287          hdfs://hadoop01:51001/hbase/statistic_visit1/7865a928112751bcf27fce2cc8304441
> 80           hdfs://hadoop01:51001/hbase/statistic_visit1/_balancedSplit
> 58234787479  hdfs://hadoop01:51001/hbase/statistic_visit1/e15d089574adaeacaf75fd8e1277ce2d
>
>     Does the split result look fine? And how can I get more regions?
>
>     And is there any danger, if I resize the region file size setting
> to a smaller value? (like 5G)
>
>     Thanks.
> --
>
> Regards,
> Cheng Su

--
Davey Yan

Hi, Davey

Thanks for your reply.

I will try the firstrow and lastrow options.

And I do get this /by zero error.
Sorry for silly question, how can I patch it?
Should I modify the source and rebuild it? or is there a compiled jar
file available?

On Wed, Nov 14, 2012 at 4:24 PM, Davey Yan <[EMAIL PROTECTED]> wrote:
> Are the options 'firstrow' and 'lastrow'  what you want?
>
> I have noticed that the '_balancedSplit' file was still there, so ...
> you must got this:
> https://issues.apache.org/jira/browse/HBASE-7138
>
> On Fri, Nov 9, 2012 at 3:47 PM, Cheng Su <[EMAIL PROTECTED]> wrote:
>> Hi, all
>>
>>     My cluster uses hbase 0.92.0.
>>     I'm now trying to split some big regions manually. (I configured
>> region file size to 20G)
>>     I ran it like this:
>> hbase org.apache.hadoop.hbase.util.RegionSplitter statistic_visit1 -r -o 2
>>
>>     Before I split, the table statistic_visit1 has only one region:
>> [hadoop@hadoop01 ~]$ hadoop dfs -du /hbase/statistic_visit1
>> Found 3 items
>> 5161         hdfs://hadoop01:51001/hbase/statistic_visit1/.tableinfo.0000000001
>> 0            hdfs://hadoop01:51001/hbase/statistic_visit1/.tmp
>> 58234791267  hdfs://hadoop01:51001/hbase/statistic_visit1/e15d089574adaeacaf75fd8e1277ce2d
>>
>>     And after the split is done,
>> [hadoop@hadoop01 ~]$ hadoop dfs -du /hbase/statistic_visit1
>> Found 6 items
>> 5161         hdfs://hadoop01:51001/hbase/statistic_visit1/.tableinfo.0000000001
>> 0            hdfs://hadoop01:51001/hbase/statistic_visit1/.tmp
>> 58234783527  hdfs://hadoop01:51001/hbase/statistic_visit1/0eae290663e59659fc728333d0b18b47
>> 287          hdfs://hadoop01:51001/hbase/statistic_visit1/7865a928112751bcf27fce2cc8304441
>> 80           hdfs://hadoop01:51001/hbase/statistic_visit1/_balancedSplit
>> 58234787479  hdfs://hadoop01:51001/hbase/statistic_visit1/e15d089574adaeacaf75fd8e1277ce2d
>>
>>     Does the split result look fine? And how can I get more regions?
>>
>>     And is there any danger, if I resize the region file size setting
>> to a smaller value? (like 5G)
>>
>>     Thanks.
>> --
>>
>> Regards,
>> Cheng Su
>
>
>
> --
> Davey Yan

--

Regards,
Cheng Su