accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Billie Rinaldi (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (ACCUMULO-1486) Hadoop 3 support
Date Mon, 10 Jun 2013 16:51:20 GMT

    [ https://issues.apache.org/jira/browse/ACCUMULO-1486?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13679626#comment-13679626
] 

Billie Rinaldi edited comment on ACCUMULO-1486 at 6/10/13 4:51 PM:
-------------------------------------------------------------------

I've gotten successful runs of most of the functional tests on hadoop 3 (my laptop stops wanting
to run them somewhere in the middle of stress.*).  I made a few minor tweaks to get this to
happen, though they don't appear to be specific to hadoop 3:

* In TestUtils.getTableId, I had to allow for the warning statement "Unable to load native-hadoop
library" since hadoop doesn't have native libs for Mac.
* In examples, the dir list example was trying to index ACCUMULO_HOME/fate, which didn't exist
due to the new packaging; I made it index the test directory instead, and search for itself
(test/system/auto/simple/examples.py).
* LargeRowTest was failing again because it found 26 split points!  I am confused because
I calculated that it should only have 16 split points (ACCUMULO-1227).  I upped the max number
of splits to NUM_PRE_SPLITS * 4 and it passed, but I did not look into why it had too many.

These fixes seem worth checking in.  I'm planning to try to run the tests for 1.5 branch (not
with hadoop 3) to see if they apply there as well.
                
      was (Author: billie.rinaldi):
    I've gotten successful runs of most of the functional tests on hadoop 3 (my laptop stops
wanting to run them somewhere in the middle of stress.*).  I made a few minor tweaks to get
this to happen, though they don't appear to be specific to hadoop 3:

* In TestUtils.getTableId, I had to allow for the warning statement "Unable to load native-hadoop
library" since hadoop doesn't have native libs for Mac.
* In examples, the dir list example was trying to index ACCUMULO_HOME/fate, which didn't exist
due to the new packaging; I made it index the test directory instead, and search for itself
(test/system/auto/simple/examples.py).
* LargeRowTest was failing again because it found 26 split points!  I am confused because
I calculated that it should only have [16 split points](ACCUMULO-1227).  I upped the max number
of splits to NUM_PRE_SPLITS * 4 and it passed, but I did not look into why it had too many.

These fixes seem worth checking in.  I'm planning to try to run the tests for 1.5 branch (not
with hadoop 3) to see if they apply there as well.
                  
> Hadoop 3 support
> ----------------
>
>                 Key: ACCUMULO-1486
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-1486
>             Project: Accumulo
>          Issue Type: Improvement
>            Reporter: Billie Rinaldi
>            Assignee: Billie Rinaldi
>         Attachments: hadoop3.patch
>
>
> Issues identified thus far:
> * sync has been removed in favor of hsync

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message