You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hbase.apache.org by "Jean-Daniel Cryans (JIRA)" <ji...@apache.org> on 2009/02/06 21:44:59 UTC
[jira] Created: (HBASE-1187) After disabling/enabling a table, the
regions seems to be assigned to only 1-2 region servers
After disabling/enabling a table, the regions seems to be assigned to only 1-2 region servers
---------------------------------------------------------------------------------------------
Key: HBASE-1187
URL: https://issues.apache.org/jira/browse/HBASE-1187
Project: Hadoop HBase
Issue Type: Bug
Components: master
Affects Versions: 0.19.0
Reporter: Jean-Daniel Cryans
After disabling/enabling a small table (20 regions), we see that the master tend to assign the regions to only 1-2 region servers. Unfortunately, that table is extensively used in random reads which really kills those RS when they hold those regions. As a fix, we have to restart HBase...
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HBASE-1187) After disabling/enabling a table,
the regions seems to be assigned to only 1-2 region servers
Posted by "Jean-Daniel Cryans (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-1187?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12671311#action_12671311 ]
Jean-Daniel Cryans commented on HBASE-1187:
-------------------------------------------
Example output from the Master after enabling the table :
{code}
2009-02-06 15:18:08,519 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entity9178d933-9237-4fe2-902c-85cad065a6e9_toponym--nothing,1233946359836 to server 192.168.1.104:62020
2009-02-06 15:18:08,525 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entityf4be6cf4-72b1-4490-9dba-5f733335c1b5_toponym--nothing,1233946367273 to server 192.168.1.104:62020
2009-02-06 15:18:08,526 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,,1233946389321 to server 192.168.1.104:62020
2009-02-06 15:18:08,527 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entity1f835c88-eb63-495b-9fcc-caa52af3b042_toponym--nothing,1233946379754 to server 192.168.1.104:62020
2009-02-06 15:18:08,527 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entity76161fb2-2c5e-4f54-aebb-219790874e2c_toponym--nothing,1233946382069 to server 192.168.1.104:62020
2009-02-06 15:18:08,528 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entitye7f6aafc-8f87-441d-886a-5c3bd82d910a_toponym--nothing,1233946365992 to server 192.168.1.104:62020
2009-02-06 15:18:08,528 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entityd6367530-88d0-48e6-8e26-4e914375035f_toponym--nothing,1233946383114 to server 192.168.1.104:62020
2009-02-06 15:18:08,529 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entity137607a0-375f-46d2-a9bd-251e9370e00c_toponym--nothing,1233946379754 to server 192.168.1.104:62020
2009-02-06 15:18:08,530 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entity97d1a639-54ef-42a8-a92a-cde538a374d7_toponym--nothing,1233946359836 to server 192.168.1.104:62020
2009-02-06 15:18:08,530 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entityb4940132-0d0b-49b6-978e-88b2ba6895bb_toponym--nothing,1233946364409 to server 192.168.1.104:62020
2009-02-06 15:18:08,976 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entitye16aa6c6-a813-4468-98c4-8e517b8b8795_toponym--nothing,1233946365992 to server 192.168.1.102:62020
2009-02-06 15:18:08,977 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entityd09c5598-2b40-4638-b3cf-daa460a50669_toponym--nothing,1233946383114 to server 192.168.1.102:62020
2009-02-06 15:18:08,978 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entity3d82e567-81ca-4fd7-b9d6-c5399cb5b32e_toponym--nothing,1233946372763 to server 192.168.1.102:62020
2009-02-06 15:18:08,978 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entityc431fc43-8ffa-4758-b888-88df542499f1_toponym--nothing,1233862204887 to server 192.168.1.102:62020
2009-02-06 15:18:08,979 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entitya9b51889-6134-46ad-b66a-85e748e66d4b_toponym--nothing,1233946364409 to server 192.168.1.102:62020
2009-02-06 15:18:08,980 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entity06281ffe-9440-4093-9948-8b9321d8a81e_toponym--nothing,1233946389321 to server 192.168.1.102:62020
2009-02-06 15:18:08,980 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entity62eec3f3-2bb2-4227-bf2e-120619d8283d_toponym--nothing,1233946382069 to server 192.168.1.102:62020
2009-02-06 15:18:08,981 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entityfb81a7ed-6a27-411f-a3cc-537e73976c42_toponym--nothing,1233946367273 to server 192.168.1.102:62020
2009-02-06 15:18:08,981 INFO org.apache.hadoop.hbase.master.RegionManager: assigning region classifiers,NED:Entity531f0d68-3621-4c14-9dda-9db0ee99bca9_toponym--nothing,1233946372763 to server 192.168.1.102:62020
{code}
> After disabling/enabling a table, the regions seems to be assigned to only 1-2 region servers
> ---------------------------------------------------------------------------------------------
>
> Key: HBASE-1187
> URL: https://issues.apache.org/jira/browse/HBASE-1187
> Project: Hadoop HBase
> Issue Type: Bug
> Components: master
> Affects Versions: 0.19.0
> Reporter: Jean-Daniel Cryans
>
> After disabling/enabling a small table (20 regions), we see that the master tend to assign the regions to only 1-2 region servers. Unfortunately, that table is extensively used in random reads which really kills those RS when they hold those regions. As a fix, we have to restart HBase...
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HBASE-1187) After disabling/enabling a table,
the regions seems to be assigned to only 1-2 region servers
Posted by "stack (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-1187?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12671522#action_12671522 ]
stack commented on HBASE-1187:
------------------------------
JD: We assign in lumps of ten. Is that issue? Is how many we assign at a time configurable?
> After disabling/enabling a table, the regions seems to be assigned to only 1-2 region servers
> ---------------------------------------------------------------------------------------------
>
> Key: HBASE-1187
> URL: https://issues.apache.org/jira/browse/HBASE-1187
> Project: Hadoop HBase
> Issue Type: Bug
> Components: master
> Affects Versions: 0.19.0
> Reporter: Jean-Daniel Cryans
>
> After disabling/enabling a small table (20 regions), we see that the master tend to assign the regions to only 1-2 region servers. Unfortunately, that table is extensively used in random reads which really kills those RS when they hold those regions. As a fix, we have to restart HBase...
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Resolved: (HBASE-1187) After disabling/enabling a table, the
regions seems to be assigned to only 1-2 region servers
Posted by "Jean-Daniel Cryans (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-1187?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jean-Daniel Cryans resolved HBASE-1187.
---------------------------------------
Resolution: Fixed
Fix Version/s: 0.20.0
0.19.1
Assignee: Jean-Daniel Cryans
Added to hbase-default.xml in branch 0.19 and trunk.
> After disabling/enabling a table, the regions seems to be assigned to only 1-2 region servers
> ---------------------------------------------------------------------------------------------
>
> Key: HBASE-1187
> URL: https://issues.apache.org/jira/browse/HBASE-1187
> Project: Hadoop HBase
> Issue Type: Bug
> Components: master
> Affects Versions: 0.19.0
> Reporter: Jean-Daniel Cryans
> Assignee: Jean-Daniel Cryans
> Fix For: 0.19.1, 0.20.0
>
>
> After disabling/enabling a small table (20 regions), we see that the master tend to assign the regions to only 1-2 region servers. Unfortunately, that table is extensively used in random reads which really kills those RS when they hold those regions. As a fix, we have to restart HBase...
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HBASE-1187) After disabling/enabling a table,
the regions seems to be assigned to only 1-2 region servers
Posted by "Jean-Daniel Cryans (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-1187?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12671658#action_12671658 ]
Jean-Daniel Cryans commented on HBASE-1187:
-------------------------------------------
I think it could be an issue for tables that stay small under particular situations. If hbase.regions.percheckin was configurable, it would solve our problem. RegionManager is even ready for this.
{code}
this.maxAssignInOneGo = conf.getInt("hbase.regions.percheckin", 10);
{code}
I will just add it to hbase-default.xml and close this case. Thank you for the pointer Stack!
> After disabling/enabling a table, the regions seems to be assigned to only 1-2 region servers
> ---------------------------------------------------------------------------------------------
>
> Key: HBASE-1187
> URL: https://issues.apache.org/jira/browse/HBASE-1187
> Project: Hadoop HBase
> Issue Type: Bug
> Components: master
> Affects Versions: 0.19.0
> Reporter: Jean-Daniel Cryans
>
> After disabling/enabling a small table (20 regions), we see that the master tend to assign the regions to only 1-2 region servers. Unfortunately, that table is extensively used in random reads which really kills those RS when they hold those regions. As a fix, we have to restart HBase...
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.