You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ambari.apache.org by Jonathan Hurley <jh...@hortonworks.com> on 2014/05/21 19:19:22 UTC

Review Request 21775: AMBARI-5840 - Ambari recomission of datanodes will not work if the active namenode is "nn2"

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/21775/
-----------------------------------------------------------

Review request for Ambari, Mahadev Konar and Sumit Mohanty.


Bugs: AMBARI-5840
    https://issues.apache.org/jira/browse/AMBARI-5840


Repository: ambari


Description
-------

hdfs doesn't update both namenodes when the refreshNodes command is issued. We are correctly executing the commands on both namenode hosts, but the bug is that only nn1 is being refreshed (twice) and nn2 is left out of it (or vice versa whichever namenode is active).

In order to fix this problem, we need to execute a scoped hdfs command that targets the namenodes explicitly. 


Diffs
-----

  ambari-server/src/main/resources/stacks/HDP/2.0.6/services/HDFS/package/scripts/hdfs_namenode.py 0086b13 
  ambari-server/src/main/resources/stacks/HDP/2.0.6/services/HDFS/package/scripts/params.py a6fea72 

Diff: https://reviews.apache.org/r/21775/diff/


Testing
-------

Results :
Tests run: 1609, Failures: 0, Errors: 0, Skipped: 13

OK
----------------------------------------------------------------------
Total run:554
Total errors:0
Total failures:0

[INFO] ------------------------------------------------------------------------
[INFO] BUILD SUCCESS
[INFO] ------------------------------------------------------------------------
[INFO] Total time: 13:06.288s
[INFO] Finished at: Wed May 21 09:46:33 PDT 2014
[INFO] Final Memory: 20M/123M
[INFO] ------------------------------------------------------------------------


Thanks,

Jonathan Hurley


Re: Review Request 21775: AMBARI-5840 - Ambari recomission of datanodes will not work if the active namenode is "nn2"

Posted by Mahadev Konar <ma...@apache.org>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/21775/#review43640
-----------------------------------------------------------


The patch looks good. JOnathan can we add some unit tests around this? Maybe mock the call via python and verify it in the unit tests? 

- Mahadev Konar


On May 21, 2014, 5:19 p.m., Jonathan Hurley wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/21775/
> -----------------------------------------------------------
> 
> (Updated May 21, 2014, 5:19 p.m.)
> 
> 
> Review request for Ambari, Mahadev Konar and Sumit Mohanty.
> 
> 
> Bugs: AMBARI-5840
>     https://issues.apache.org/jira/browse/AMBARI-5840
> 
> 
> Repository: ambari
> 
> 
> Description
> -------
> 
> hdfs doesn't update both namenodes when the refreshNodes command is issued. We are correctly executing the commands on both namenode hosts, but the bug is that only nn1 is being refreshed (twice) and nn2 is left out of it (or vice versa whichever namenode is active).
> 
> In order to fix this problem, we need to execute a scoped hdfs command that targets the namenodes explicitly. 
> 
> 
> Diffs
> -----
> 
>   ambari-server/src/main/resources/stacks/HDP/2.0.6/services/HDFS/package/scripts/hdfs_namenode.py 0086b13 
>   ambari-server/src/main/resources/stacks/HDP/2.0.6/services/HDFS/package/scripts/params.py a6fea72 
> 
> Diff: https://reviews.apache.org/r/21775/diff/
> 
> 
> Testing
> -------
> 
> Results :
> Tests run: 1609, Failures: 0, Errors: 0, Skipped: 13
> 
> OK
> ----------------------------------------------------------------------
> Total run:554
> Total errors:0
> Total failures:0
> 
> [INFO] ------------------------------------------------------------------------
> [INFO] BUILD SUCCESS
> [INFO] ------------------------------------------------------------------------
> [INFO] Total time: 13:06.288s
> [INFO] Finished at: Wed May 21 09:46:33 PDT 2014
> [INFO] Final Memory: 20M/123M
> [INFO] ------------------------------------------------------------------------
> 
> 
> Thanks,
> 
> Jonathan Hurley
> 
>


Re: Review Request 21775: AMBARI-5840 - Ambari recomission of datanodes will not work if the active namenode is "nn2"

Posted by Mahadev Konar <ma...@apache.org>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/21775/#review43735
-----------------------------------------------------------

Ship it!


Ship It!

- Mahadev Konar


On May 21, 2014, 10:49 p.m., Jonathan Hurley wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/21775/
> -----------------------------------------------------------
> 
> (Updated May 21, 2014, 10:49 p.m.)
> 
> 
> Review request for Ambari, Mahadev Konar and Sumit Mohanty.
> 
> 
> Bugs: AMBARI-5840
>     https://issues.apache.org/jira/browse/AMBARI-5840
> 
> 
> Repository: ambari
> 
> 
> Description
> -------
> 
> hdfs doesn't update both namenodes when the refreshNodes command is issued. We are correctly executing the commands on both namenode hosts, but the bug is that only nn1 is being refreshed (twice) and nn2 is left out of it (or vice versa whichever namenode is active).
> 
> In order to fix this problem, we need to execute a scoped hdfs command that targets the namenodes explicitly. 
> 
> 
> Diffs
> -----
> 
>   ambari-server/src/main/resources/stacks/HDP/2.0.6/services/HDFS/package/scripts/hdfs_namenode.py 0086b13 
>   ambari-server/src/main/resources/stacks/HDP/2.0.6/services/HDFS/package/scripts/params.py a6fea72 
>   ambari-server/src/test/python/stacks/2.0.6/HDFS/test_namenode.py 54d2175 
> 
> Diff: https://reviews.apache.org/r/21775/diff/
> 
> 
> Testing
> -------
> 
> Results :
> Tests run: 1609, Failures: 0, Errors: 0, Skipped: 13
> 
> OK
> ----------------------------------------------------------------------
> Total run:555
> Total errors:0
> Total failures:0
> 
> [INFO] ------------------------------------------------------------------------
> [INFO] BUILD SUCCESS
> [INFO] ------------------------------------------------------------------------
> [INFO] Total time: 13:06.288s
> [INFO] Finished at: Wed May 21 09:46:33 PDT 2014
> [INFO] Final Memory: 20M/123M
> [INFO] ------------------------------------------------------------------------
> 
> 
> Thanks,
> 
> Jonathan Hurley
> 
>


Re: Review Request 21775: AMBARI-5840 - Ambari recomission of datanodes will not work if the active namenode is "nn2"

Posted by Jonathan Hurley <jh...@hortonworks.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/21775/
-----------------------------------------------------------

(Updated May 21, 2014, 6:49 p.m.)


Review request for Ambari, Mahadev Konar and Sumit Mohanty.


Changes
-------

Added unit test to cover calling hdfs -fs on nn1 in HA mode.


Bugs: AMBARI-5840
    https://issues.apache.org/jira/browse/AMBARI-5840


Repository: ambari


Description
-------

hdfs doesn't update both namenodes when the refreshNodes command is issued. We are correctly executing the commands on both namenode hosts, but the bug is that only nn1 is being refreshed (twice) and nn2 is left out of it (or vice versa whichever namenode is active).

In order to fix this problem, we need to execute a scoped hdfs command that targets the namenodes explicitly. 


Diffs (updated)
-----

  ambari-server/src/main/resources/stacks/HDP/2.0.6/services/HDFS/package/scripts/hdfs_namenode.py 0086b13 
  ambari-server/src/main/resources/stacks/HDP/2.0.6/services/HDFS/package/scripts/params.py a6fea72 
  ambari-server/src/test/python/stacks/2.0.6/HDFS/test_namenode.py 54d2175 

Diff: https://reviews.apache.org/r/21775/diff/


Testing (updated)
-------

Results :
Tests run: 1609, Failures: 0, Errors: 0, Skipped: 13

OK
----------------------------------------------------------------------
Total run:555
Total errors:0
Total failures:0

[INFO] ------------------------------------------------------------------------
[INFO] BUILD SUCCESS
[INFO] ------------------------------------------------------------------------
[INFO] Total time: 13:06.288s
[INFO] Finished at: Wed May 21 09:46:33 PDT 2014
[INFO] Final Memory: 20M/123M
[INFO] ------------------------------------------------------------------------


Thanks,

Jonathan Hurley