Hello,

I created yesterday an HTable with 2 CF specifying the TTL for 5 an 10
min respectively.

Inserted 2 datas (one in each column)

And hoped that my values desapear passed a certain amount of time.

This never happend ...

This morning I keep hope that major_compaction once a days withdraw my
expired datas ...

Disapointed :

hbase(main):004:0> describe 'ttltest'
DESCRIPTION                                                                                                                          
ENABLED
  {NAME => 'ttltest', FAMILIES => [{NAME => 'longttl', BLOOMFILTER =>
'NONE', REPLICATION_SCOPE => '0', COMPRESSION => 'NONE', VERSION true
  S => '1', TTL => '600000', BLOCKSIZE => '65536', IN_MEMORY => 'false',
BLOCKCACHE => 'true'}, {NAME => 'shortttl', BLOOMFILTER => 'N
  ONE', REPLICATION_SCOPE => '0', COMPRESSION => 'NONE', VERSIONS =>
'1', TTL => '300000', BLOCKSIZE => '65536', IN_MEMORY => 'false',
   BLOCKCACHE => 'true'}]}

hbase(main):005:0> scan 'ttltest'
ROW                                                  COLUMN+CELL
  test1                                              
column=longttl:data, timestamp=1316697813182, value=jeudi 22 septembre
2011, 15:22:06 (UTC+0200)
  test1                                              
column=shortttl:data, timestamp=1316697803522, value=jeudi 22 septembre
2011, 15:22:06 (UTC+0200)
1 row(s) in 0.1280 seconds
Is TTL usable to purge expired values (even the last ones) ?

Best regards,

--
Damien

Did major compaction run?  You could manually run it from the shell.

Check the logs to see it actually completed.   If it did, and the
values are still in place, then file a JIRA.  It sounds like a bug.

St.Ack

On Fri, Sep 23, 2011 at 2:22 AM, Damien Hardy <[EMAIL PROTECTED]> wrote:
> Hello,
>
> I created yesterday an HTable with 2 CF specifying the TTL for 5 an 10 min
> respectively.
>
> Inserted 2 datas (one in each column)
>
> And hoped that my values desapear passed a certain amount of time.
>
> This never happend ...
>
> This morning I keep hope that major_compaction once a days withdraw my
> expired datas ...
>
> Disapointed :
>
> hbase(main):004:0> describe 'ttltest'
> DESCRIPTION
>                                                           ENABLED
>  {NAME => 'ttltest', FAMILIES => [{NAME => 'longttl', BLOOMFILTER => 'NONE',
> REPLICATION_SCOPE => '0', COMPRESSION => 'NONE', VERSION true
>  S => '1', TTL => '600000', BLOCKSIZE => '65536', IN_MEMORY => 'false',
> BLOCKCACHE => 'true'}, {NAME => 'shortttl', BLOOMFILTER => 'N
>  ONE', REPLICATION_SCOPE => '0', COMPRESSION => 'NONE', VERSIONS => '1', TTL
> => '300000', BLOCKSIZE => '65536', IN_MEMORY => 'false',
>  BLOCKCACHE => 'true'}]}
>
> hbase(main):005:0> scan 'ttltest'
> ROW                                                  COLUMN+CELL
>  test1                                               column=longttl:data,
> timestamp=1316697813182, value=jeudi 22 septembre 2011, 15:22:06 (UTC+0200)
>  test1                                               column=shortttl:data,
> timestamp=1316697803522, value=jeudi 22 septembre 2011, 15:22:06 (UTC+0200)
> 1 row(s) in 0.1280 seconds
>
>
> Is TTL usable to purge expired values (even the last ones) ?
>
> Best regards,
>
> --
> Damien
>

>From the book http://hbase.apache.org/book/ttl.html

"ColumnFamilies can set a TTL length in seconds"

and you have:

 TTL => '600000'
 TTL => '300000'

It's just three orders of magnitude different from what you thought
you set the TTL to :)

J-D

On Fri, Sep 23, 2011 at 2:22 AM, Damien Hardy <[EMAIL PROTECTED]> wrote:
> Hello,
>
> I created yesterday an HTable with 2 CF specifying the TTL for 5 an 10 min
> respectively.
>
> Inserted 2 datas (one in each column)
>
> And hoped that my values desapear passed a certain amount of time.
>
> This never happend ...
>
> This morning I keep hope that major_compaction once a days withdraw my
> expired datas ...
>
> Disapointed :
>
> hbase(main):004:0> describe 'ttltest'
> DESCRIPTION
>                                                           ENABLED
>  {NAME => 'ttltest', FAMILIES => [{NAME => 'longttl', BLOOMFILTER => 'NONE',
> REPLICATION_SCOPE => '0', COMPRESSION => 'NONE', VERSION true
>  S => '1', TTL => '600000', BLOCKSIZE => '65536', IN_MEMORY => 'false',
> BLOCKCACHE => 'true'}, {NAME => 'shortttl', BLOOMFILTER => 'N
>  ONE', REPLICATION_SCOPE => '0', COMPRESSION => 'NONE', VERSIONS => '1', TTL
> => '300000', BLOCKSIZE => '65536', IN_MEMORY => 'false',
>  BLOCKCACHE => 'true'}]}
>
> hbase(main):005:0> scan 'ttltest'
> ROW                                                  COLUMN+CELL
>  test1                                               column=longttl:data,
> timestamp=1316697813182, value=jeudi 22 septembre 2011, 15:22:06 (UTC+0200)
>  test1                                               column=shortttl:data,
> timestamp=1316697803522, value=jeudi 22 septembre 2011, 15:22:06 (UTC+0200)
> 1 row(s) in 0.1280 seconds
>
>
> Is TTL usable to purge expired values (even the last ones) ?
>
> Best regards,
>
> --
> Damien
>

This is an occasional source of confusion.

Curious if anyone thinks that TTLs in milliseconds makes sense.

My opinion is no, of what practical use is data with a lifetime of 1, 10, or 100 ms?
----- Original Message -----
> From: Jean-Daniel Cryans <[EMAIL PROTECTED]>
> To: [EMAIL PROTECTED]
> Cc:
> Sent: Friday, September 23, 2011 9:40 AM
> Subject: Re: Using TTL tout purge data automatically ?
>
> From the book http://hbase.apache.org/book/ttl.html
>
> "ColumnFamilies can set a TTL length in seconds"
>
> and you have:
>
> TTL => '600000'
> TTL => '300000'
>
> It's just three orders of magnitude different from what you thought
> you set the TTL to :)
>
> J-D
>
> On Fri, Sep 23, 2011 at 2:22 AM, Damien Hardy <[EMAIL PROTECTED]> wrote:
>>  Hello,
>>
>>  I created yesterday an HTable with 2 CF specifying the TTL for 5 an 10 min
>>  respectively.
>>
>>  Inserted 2 datas (one in each column)
>>
>>  And hoped that my values desapear passed a certain amount of time.
>>
>>  This never happend ...
>>
>>  This morning I keep hope that major_compaction once a days withdraw my
>>  expired datas ...
>>
>>  Disapointed :
>>
>>  hbase(main):004:0> describe 'ttltest'
>>  DESCRIPTION
>>                                                            ENABLED
>>   {NAME => 'ttltest', FAMILIES => [{NAME =>
> 'longttl', BLOOMFILTER => 'NONE',
>>  REPLICATION_SCOPE => '0', COMPRESSION => 'NONE',
> VERSION true
>>   S => '1', TTL => '600000', BLOCKSIZE =>
> '65536', IN_MEMORY => 'false',
>>  BLOCKCACHE => 'true'}, {NAME => 'shortttl',
> BLOOMFILTER => 'N
>>   ONE', REPLICATION_SCOPE => '0', COMPRESSION =>
> 'NONE', VERSIONS => '1', TTL
>>  => '300000', BLOCKSIZE => '65536', IN_MEMORY =>
> 'false',
>>   BLOCKCACHE => 'true'}]}
>>
>>  hbase(main):005:0> scan 'ttltest'
>>  ROW                                                  COLUMN+CELL
>>   test1                                               column=longttl:data,
>>  timestamp=1316697813182, value=jeudi 22 septembre 2011, 15:22:06 (UTC+0200)
>>   test1                                               column=shortttl:data,
>>  timestamp=1316697803522, value=jeudi 22 septembre 2011, 15:22:06 (UTC+0200)
>>  1 row(s) in 0.1280 seconds
>>
>>
>>  Is TTL usable to purge expired values (even the last ones) ?
>>
>>  Best regards,
>>
>>  --
>>  Damien
>>
>

Thank-you for your answer actually i suppose i was confused by the timestamp unit of the values in ms.
I'll try with properer values on Monday.

Envoyé depuis mon HTC

----- Reply message -----
De : "Andrew Purtell" <[EMAIL PROTECTED]>
Date : ven., sept. 23, 2011 18:50
Objet : Using TTL tout purge data automatically ?
Pour : "[EMAIL PROTECTED]" <[EMAIL PROTECTED]>

This is an occasional source of confusion.

Curious if anyone thinks that TTLs in milliseconds makes sense.

My opinion is no, of what practical use is data with a lifetime of 1, 10, or 100 ms?
----- Original Message -----
> From: Jean-Daniel Cryans <[EMAIL PROTECTED]>
> To: [EMAIL PROTECTED]
> Cc:
> Sent: Friday, September 23, 2011 9:40 AM
> Subject: Re: Using TTL tout purge data automatically ?
>
> From the book http://hbase.apache.org/book/ttl.html
>
> "ColumnFamilies can set a TTL length in seconds"
>
> and you have:
>
> TTL => '600000'
> TTL => '300000'
>
> It's just three orders of magnitude different from what you thought
> you set the TTL to :)
>
> J-D
>
> On Fri, Sep 23, 2011 at 2:22 AM, Damien Hardy <[EMAIL PROTECTED]> wrote:
>>  Hello,
>>
>>  I created yesterday an HTable with 2 CF specifying the TTL for 5 an 10 min
>>  respectively.
>>
>>  Inserted 2 datas (one in each column)
>>
>>  And hoped that my values desapear passed a certain amount of time.
>>
>>  This never happend ...
>>
>>  This morning I keep hope that major_compaction once a days withdraw my
>>  expired datas ...
>>
>>  Disapointed :
>>
>>  hbase(main):004:0> describe 'ttltest'
>>  DESCRIPTION
>>                                                            ENABLED
>>   {NAME => 'ttltest', FAMILIES => [{NAME =>
> 'longttl', BLOOMFILTER => 'NONE',
>>  REPLICATION_SCOPE => '0', COMPRESSION => 'NONE',
> VERSION true
>>   S => '1', TTL => '600000', BLOCKSIZE =>
> '65536', IN_MEMORY => 'false',
>>  BLOCKCACHE => 'true'}, {NAME => 'shortttl',
> BLOOMFILTER => 'N
>>   ONE', REPLICATION_SCOPE => '0', COMPRESSION =>
> 'NONE', VERSIONS => '1', TTL
>>  => '300000', BLOCKSIZE => '65536', IN_MEMORY =>
> 'false',
>>   BLOCKCACHE => 'true'}]}
>>
>>  hbase(main):005:0> scan 'ttltest'
>>  ROW                                                  COLUMN+CELL
>>   test1                                               column=longttl:data,
>>  timestamp=1316697813182, value=jeudi 22 septembre 2011, 15:22:06 (UTC+0200)
>>   test1                                               column=shortttl:data,
>>  timestamp=1316697803522, value=jeudi 22 septembre 2011, 15:22:06 (UTC+0200)
>>  1 row(s) in 0.1280 seconds
>>
>>
>>  Is TTL usable to purge expired values (even the last ones) ?
>>
>>  Best regards,
>>
>>  --
>>  Damien
>>
>