You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@hbase.apache.org by Serega Sheypak <se...@gmail.com> on 2014/10/04 14:59:41 UTC
weird problem when creating a table using hbulk loading to table.se shell
Hi, I've created a table
create 'my_table', {NAME => 'cf', VERSIONS => '1', IN_MEMORY => true,
BLOCKCACHE => true, BLOOMFILTER => 'ROWCOL', SPLITS =>
["\x00\x00\x00\x12d\xF47\x9F", "#255 more keys"]}
then I ttry to descrbe it:
hbase(main):018:0> describe 'my_table'
DESCRIPTION
ENABLED
'my_table'
true
1 row(s) in 0.0300 seconds
And usual output is:
'some_other_table', {NAME => 'ctx', DATA_BLOCK_ENCODING => 'NONE',
BLOOMFILTER => 'NONE', REPLICATION_SCOPE => '0', VERSION true
S => '1', COMPRESSION => 'SNAPPY', MIN_VERSIONS => '0', TTL =>
'2147483647', KEEP_DELETED_CELLS => 'false', BLOCKSIZE => '655
36', IN_MEMORY => 'false', ENCODE_ON_DISK => 'true', BLOCKCACHE => 'true'}
1 row(s) in 0.0320 seconds
Re: weird problem when creating a table using hbulk loading to
table.se shell
Posted by Serega Sheypak <se...@gmail.com>.
> there was syntax error: SPLITS
thanks!
What's subscriber_network ?
subscriber_network = my_table
Can you use Java API to create the table with the 256 splits ?
Already did it, works!
2014-10-04 19:56 GMT+04:00 Ted Yu <yu...@gmail.com>:
> In the shell command from first email, there was syntax error: SPLITS
> should be for table, not column family 'cf'.
>
> What's subscriber_network ?
>
> Can you use Java API to create the table with the 256 splits ?
>
> Cheers
>
> On Sat, Oct 4, 2014 at 8:44 AM, Serega Sheypak <se...@gmail.com>
> wrote:
>
> > I've found this one:
> >
> > Failed getting descriptor for subscriber_network
> > java.io.FileNotFoundException: File does not exist:
> > hdfs://nameservice1/hbase/my_table
> > at
> >
> org.apache.hadoop.hdfs.DistributedFileSystem.getFileStatus(DistributedFileSystem.java:824)
> >
> >
> > And then tons of:
> >
> > creating HRegion subscriber_network HTD == 'my_table' RootDir =
> > hdfs://nameservice1/hbase/.tmp Table name == my_table
> >
> >
> > This problem is related to splits. If I remove split SPLIT=>[k1, k2,
> > ...] declarartion from table DDL, table appears normally but without
> > splits ofcourse
> >
> >
> >
> >
> > 2014-10-04 18:01 GMT+04:00 Ted Yu <yu...@gmail.com>:
> >
> > > Can you use Java API to retrieve table description ?
> > >
> > > Also take a look at master log to see if there is any error /
> exception.
> > >
> > > Cheers
> > >
> > > On Oct 4, 2014, at 5:59 AM, Serega Sheypak <se...@gmail.com>
> > > wrote:
> > >
> > > > Hi, I've created a table
> > > >
> > > > create 'my_table', {NAME => 'cf', VERSIONS => '1', IN_MEMORY => true,
> > > > BLOCKCACHE => true, BLOOMFILTER => 'ROWCOL', SPLITS =>
> > > > ["\x00\x00\x00\x12d\xF47\x9F", "#255 more keys"]}
> > > >
> > > > then I ttry to descrbe it:
> > > >
> > > > hbase(main):018:0> describe 'my_table'
> > > > DESCRIPTION
> > > > ENABLED
> > > >
> > > > 'my_table'
> > > > true
> > > >
> > > > 1 row(s) in 0.0300 seconds
> > > >
> > > > And usual output is:
> > > >
> > > > 'some_other_table', {NAME => 'ctx', DATA_BLOCK_ENCODING => 'NONE',
> > > > BLOOMFILTER => 'NONE', REPLICATION_SCOPE => '0', VERSION true
> > > >
> > > > S => '1', COMPRESSION => 'SNAPPY', MIN_VERSIONS => '0', TTL =>
> > > > '2147483647', KEEP_DELETED_CELLS => 'false', BLOCKSIZE => '655
> > > >
> > > > 36', IN_MEMORY => 'false', ENCODE_ON_DISK => 'true', BLOCKCACHE =>
> > > 'true'}
> > > >
> > > >
> > > > 1 row(s) in 0.0320 seconds
> > >
> >
>
Re: weird problem when creating a table using hbulk loading to
table.se shell
Posted by Ted Yu <yu...@gmail.com>.
In the shell command from first email, there was syntax error: SPLITS
should be for table, not column family 'cf'.
What's subscriber_network ?
Can you use Java API to create the table with the 256 splits ?
Cheers
On Sat, Oct 4, 2014 at 8:44 AM, Serega Sheypak <se...@gmail.com>
wrote:
> I've found this one:
>
> Failed getting descriptor for subscriber_network
> java.io.FileNotFoundException: File does not exist:
> hdfs://nameservice1/hbase/my_table
> at
> org.apache.hadoop.hdfs.DistributedFileSystem.getFileStatus(DistributedFileSystem.java:824)
>
>
> And then tons of:
>
> creating HRegion subscriber_network HTD == 'my_table' RootDir =
> hdfs://nameservice1/hbase/.tmp Table name == my_table
>
>
> This problem is related to splits. If I remove split SPLIT=>[k1, k2,
> ...] declarartion from table DDL, table appears normally but without
> splits ofcourse
>
>
>
>
> 2014-10-04 18:01 GMT+04:00 Ted Yu <yu...@gmail.com>:
>
> > Can you use Java API to retrieve table description ?
> >
> > Also take a look at master log to see if there is any error / exception.
> >
> > Cheers
> >
> > On Oct 4, 2014, at 5:59 AM, Serega Sheypak <se...@gmail.com>
> > wrote:
> >
> > > Hi, I've created a table
> > >
> > > create 'my_table', {NAME => 'cf', VERSIONS => '1', IN_MEMORY => true,
> > > BLOCKCACHE => true, BLOOMFILTER => 'ROWCOL', SPLITS =>
> > > ["\x00\x00\x00\x12d\xF47\x9F", "#255 more keys"]}
> > >
> > > then I ttry to descrbe it:
> > >
> > > hbase(main):018:0> describe 'my_table'
> > > DESCRIPTION
> > > ENABLED
> > >
> > > 'my_table'
> > > true
> > >
> > > 1 row(s) in 0.0300 seconds
> > >
> > > And usual output is:
> > >
> > > 'some_other_table', {NAME => 'ctx', DATA_BLOCK_ENCODING => 'NONE',
> > > BLOOMFILTER => 'NONE', REPLICATION_SCOPE => '0', VERSION true
> > >
> > > S => '1', COMPRESSION => 'SNAPPY', MIN_VERSIONS => '0', TTL =>
> > > '2147483647', KEEP_DELETED_CELLS => 'false', BLOCKSIZE => '655
> > >
> > > 36', IN_MEMORY => 'false', ENCODE_ON_DISK => 'true', BLOCKCACHE =>
> > 'true'}
> > >
> > >
> > > 1 row(s) in 0.0320 seconds
> >
>
Re: weird problem when creating a table using hbulk loading to
table.se shell
Posted by Serega Sheypak <se...@gmail.com>.
I've found this one:
Failed getting descriptor for subscriber_network
java.io.FileNotFoundException: File does not exist:
hdfs://nameservice1/hbase/my_table
at org.apache.hadoop.hdfs.DistributedFileSystem.getFileStatus(DistributedFileSystem.java:824)
And then tons of:
creating HRegion subscriber_network HTD == 'my_table' RootDir =
hdfs://nameservice1/hbase/.tmp Table name == my_table
This problem is related to splits. If I remove split SPLIT=>[k1, k2,
...] declarartion from table DDL, table appears normally but without
splits ofcourse
2014-10-04 18:01 GMT+04:00 Ted Yu <yu...@gmail.com>:
> Can you use Java API to retrieve table description ?
>
> Also take a look at master log to see if there is any error / exception.
>
> Cheers
>
> On Oct 4, 2014, at 5:59 AM, Serega Sheypak <se...@gmail.com>
> wrote:
>
> > Hi, I've created a table
> >
> > create 'my_table', {NAME => 'cf', VERSIONS => '1', IN_MEMORY => true,
> > BLOCKCACHE => true, BLOOMFILTER => 'ROWCOL', SPLITS =>
> > ["\x00\x00\x00\x12d\xF47\x9F", "#255 more keys"]}
> >
> > then I ttry to descrbe it:
> >
> > hbase(main):018:0> describe 'my_table'
> > DESCRIPTION
> > ENABLED
> >
> > 'my_table'
> > true
> >
> > 1 row(s) in 0.0300 seconds
> >
> > And usual output is:
> >
> > 'some_other_table', {NAME => 'ctx', DATA_BLOCK_ENCODING => 'NONE',
> > BLOOMFILTER => 'NONE', REPLICATION_SCOPE => '0', VERSION true
> >
> > S => '1', COMPRESSION => 'SNAPPY', MIN_VERSIONS => '0', TTL =>
> > '2147483647', KEEP_DELETED_CELLS => 'false', BLOCKSIZE => '655
> >
> > 36', IN_MEMORY => 'false', ENCODE_ON_DISK => 'true', BLOCKCACHE =>
> 'true'}
> >
> >
> > 1 row(s) in 0.0320 seconds
>
Re: weird problem when creating a table using hbulk loading to table.se shell
Posted by Ted Yu <yu...@gmail.com>.
Can you use Java API to retrieve table description ?
Also take a look at master log to see if there is any error / exception.
Cheers
On Oct 4, 2014, at 5:59 AM, Serega Sheypak <se...@gmail.com> wrote:
> Hi, I've created a table
>
> create 'my_table', {NAME => 'cf', VERSIONS => '1', IN_MEMORY => true,
> BLOCKCACHE => true, BLOOMFILTER => 'ROWCOL', SPLITS =>
> ["\x00\x00\x00\x12d\xF47\x9F", "#255 more keys"]}
>
> then I ttry to descrbe it:
>
> hbase(main):018:0> describe 'my_table'
> DESCRIPTION
> ENABLED
>
> 'my_table'
> true
>
> 1 row(s) in 0.0300 seconds
>
> And usual output is:
>
> 'some_other_table', {NAME => 'ctx', DATA_BLOCK_ENCODING => 'NONE',
> BLOOMFILTER => 'NONE', REPLICATION_SCOPE => '0', VERSION true
>
> S => '1', COMPRESSION => 'SNAPPY', MIN_VERSIONS => '0', TTL =>
> '2147483647', KEEP_DELETED_CELLS => 'false', BLOCKSIZE => '655
>
> 36', IN_MEMORY => 'false', ENCODE_ON_DISK => 'true', BLOCKCACHE => 'true'}
>
>
> 1 row(s) in 0.0320 seconds