You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-issues@hadoop.apache.org by "Jakob Homan (JIRA)" <ji...@apache.org> on 2009/07/22 01:55:14 UTC

[jira] Created: (HADOOP-6164) Test-patch's method of checking for new tests is not correct

Test-patch's method of checking for new tests is not correct
------------------------------------------------------------

                 Key: HADOOP-6164
                 URL: https://issues.apache.org/jira/browse/HADOOP-6164
             Project: Hadoop Common
          Issue Type: Bug
          Components: test
            Reporter: Jakob Homan


As happened in HDFS-458, test-patch/hudson saw the previous patch(HDFS-492) add a new test, bringing the total number of tests to 48.  Then it tested 458, which had no new tests as it was a change to the build file.  492 had not yet been committed yet, so there were still only 48 tests in trunk.  But test-patch failed the patch, expecting 49 tests.  test-patch should check the correct number of tests based on trunk, not on whatever number it last saw.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Commented: (HADOOP-6164) Test-patch's method of checking for new tests is not correct

Posted by "Giridharan Kesavan (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/HADOOP-6164?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12734543#action_12734543 ] 

Giridharan Kesavan commented on HADOOP-6164:
--------------------------------------------

I agree with you and I would follow up with this issue. 
But this is for the first time I'm hearing something like this.

If you agree that this not something to do with test-patch then I can openup a different jira and deal with hudson issue or we can file a bug with hudson dev.

thanks

> Test-patch's method of checking for new tests is not correct
> ------------------------------------------------------------
>
>                 Key: HADOOP-6164
>                 URL: https://issues.apache.org/jira/browse/HADOOP-6164
>             Project: Hadoop Common
>          Issue Type: Bug
>          Components: test
>            Reporter: Jakob Homan
>            Assignee: Giridharan Kesavan
>
> As happened in HDFS-458, test-patch/hudson saw the previous patch(HDFS-492) add a new test, bringing the total number of tests to 49.  Then it tested 458, which had no new tests as it was a change to the build file.  492 had not yet been committed yet, so there were still only 48 tests in trunk.  But test-patch failed the patch, expecting 49 tests.  test-patch should check the correct number of tests based on trunk, not on whatever number it last saw.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Commented: (HADOOP-6164) Test-patch's method of checking for new tests is not correct

Posted by "Tsz Wo (Nicholas), SZE (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/HADOOP-6164?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12734208#action_12734208 ] 

Tsz Wo (Nicholas), SZE commented on HADOOP-6164:
------------------------------------------------

> BTW hudson didnt run any core-test for HDFS-450
Why it is related to BTW HDFS-450?  Typo?

I think Jakob was talking about [this test results|https://issues.apache.org/jira/browse/HDFS-458?focusedCommentId=12732863&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#action_12732863].  It got +1 on contrib test.

Giridharan, could you take a second look?

> Test-patch's method of checking for new tests is not correct
> ------------------------------------------------------------
>
>                 Key: HADOOP-6164
>                 URL: https://issues.apache.org/jira/browse/HADOOP-6164
>             Project: Hadoop Common
>          Issue Type: Bug
>          Components: test
>            Reporter: Jakob Homan
>            Assignee: Giridharan Kesavan
>
> As happened in HDFS-458, test-patch/hudson saw the previous patch(HDFS-492) add a new test, bringing the total number of tests to 49.  Then it tested 458, which had no new tests as it was a change to the build file.  492 had not yet been committed yet, so there were still only 48 tests in trunk.  But test-patch failed the patch, expecting 49 tests.  test-patch should check the correct number of tests based on trunk, not on whatever number it last saw.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Commented: (HADOOP-6164) Test-patch's method of checking for new tests is not correct

Posted by "Giridharan Kesavan (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/HADOOP-6164?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12734905#action_12734905 ] 

Giridharan Kesavan commented on HADOOP-6164:
--------------------------------------------

filed a jira for trackin 
https://issues.apache.org/jira/browse/HDFS-502
tnx!


> Test-patch's method of checking for new tests is not correct
> ------------------------------------------------------------
>
>                 Key: HADOOP-6164
>                 URL: https://issues.apache.org/jira/browse/HADOOP-6164
>             Project: Hadoop Common
>          Issue Type: Bug
>          Components: test
>            Reporter: Jakob Homan
>            Assignee: Giridharan Kesavan
>
> As happened in HDFS-458, test-patch/hudson saw the previous patch(HDFS-492) add a new test, bringing the total number of tests to 49.  Then it tested 458, which had no new tests as it was a change to the build file.  492 had not yet been committed yet, so there were still only 48 tests in trunk.  But test-patch failed the patch, expecting 49 tests.  test-patch should check the correct number of tests based on trunk, not on whatever number it last saw.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Commented: (HADOOP-6164) Test-patch's method of checking for new tests is not correct

Posted by "Jakob Homan (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/HADOOP-6164?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12734212#action_12734212 ] 

Jakob Homan commented on HADOOP-6164:
-------------------------------------

I've spoken with Giri about this offline.  Hopefully he'll re-open the issue soon.

> Test-patch's method of checking for new tests is not correct
> ------------------------------------------------------------
>
>                 Key: HADOOP-6164
>                 URL: https://issues.apache.org/jira/browse/HADOOP-6164
>             Project: Hadoop Common
>          Issue Type: Bug
>          Components: test
>            Reporter: Jakob Homan
>            Assignee: Giridharan Kesavan
>
> As happened in HDFS-458, test-patch/hudson saw the previous patch(HDFS-492) add a new test, bringing the total number of tests to 49.  Then it tested 458, which had no new tests as it was a change to the build file.  492 had not yet been committed yet, so there were still only 48 tests in trunk.  But test-patch failed the patch, expecting 49 tests.  test-patch should check the correct number of tests based on trunk, not on whatever number it last saw.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Resolved: (HADOOP-6164) Test-patch's method of checking for new tests is not correct

Posted by "Giridharan Kesavan (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/HADOOP-6164?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Giridharan Kesavan resolved HADOOP-6164.
----------------------------------------

    Resolution: Invalid
      Assignee: Giridharan Kesavan

Im closing this resolved invalid.
This is how hudson works and this has nothing to do with the test-patch script.

> Test-patch's method of checking for new tests is not correct
> ------------------------------------------------------------
>
>                 Key: HADOOP-6164
>                 URL: https://issues.apache.org/jira/browse/HADOOP-6164
>             Project: Hadoop Common
>          Issue Type: Bug
>          Components: test
>            Reporter: Jakob Homan
>            Assignee: Giridharan Kesavan
>
> As happened in HDFS-458, test-patch/hudson saw the previous patch(HDFS-492) add a new test, bringing the total number of tests to 49.  Then it tested 458, which had no new tests as it was a change to the build file.  492 had not yet been committed yet, so there were still only 48 tests in trunk.  But test-patch failed the patch, expecting 49 tests.  test-patch should check the correct number of tests based on trunk, not on whatever number it last saw.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Commented: (HADOOP-6164) Test-patch's method of checking for new tests is not correct

Posted by "Giridharan Kesavan (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/HADOOP-6164?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12734003#action_12734003 ] 

Giridharan Kesavan commented on HADOOP-6164:
--------------------------------------------

test patch doesn make any decision on the no of tests run. If the tests fail it fails the build. It is hudson which parses the junit test results from the xml file.

BTW hudson didnt run any core-test for HDFS-450
it failed for create-configure-c++ so didnt run test-core target
so -1 for core test
and hdfsproxy test failed so -1 for contrib test.

test-patch script doesnt do any count on the no of test run.



> Test-patch's method of checking for new tests is not correct
> ------------------------------------------------------------
>
>                 Key: HADOOP-6164
>                 URL: https://issues.apache.org/jira/browse/HADOOP-6164
>             Project: Hadoop Common
>          Issue Type: Bug
>          Components: test
>            Reporter: Jakob Homan
>
> As happened in HDFS-458, test-patch/hudson saw the previous patch(HDFS-492) add a new test, bringing the total number of tests to 49.  Then it tested 458, which had no new tests as it was a change to the build file.  492 had not yet been committed yet, so there were still only 48 tests in trunk.  But test-patch failed the patch, expecting 49 tests.  test-patch should check the correct number of tests based on trunk, not on whatever number it last saw.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Commented: (HADOOP-6164) Test-patch's method of checking for new tests is not correct

Posted by "Tsz Wo (Nicholas), SZE (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/HADOOP-6164?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12734209#action_12734209 ] 

Tsz Wo (Nicholas), SZE commented on HADOOP-6164:
------------------------------------------------

> Why it is related to BTW HDFS-450? Typo? 
Oops, I got a typo. It should be "Why it is related to HDFS-450? Typo? "

> Test-patch's method of checking for new tests is not correct
> ------------------------------------------------------------
>
>                 Key: HADOOP-6164
>                 URL: https://issues.apache.org/jira/browse/HADOOP-6164
>             Project: Hadoop Common
>          Issue Type: Bug
>          Components: test
>            Reporter: Jakob Homan
>            Assignee: Giridharan Kesavan
>
> As happened in HDFS-458, test-patch/hudson saw the previous patch(HDFS-492) add a new test, bringing the total number of tests to 49.  Then it tested 458, which had no new tests as it was a change to the build file.  492 had not yet been committed yet, so there were still only 48 tests in trunk.  But test-patch failed the patch, expecting 49 tests.  test-patch should check the correct number of tests based on trunk, not on whatever number it last saw.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Commented: (HADOOP-6164) Test-patch's method of checking for new tests is not correct

Posted by "Jakob Homan (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/HADOOP-6164?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12734509#action_12734509 ] 

Jakob Homan commented on HADOOP-6164:
-------------------------------------

Well I imagine it's worth finding out why Hudson let through a patch that failed a test without reporting it.  This is a pretty big issue if it's happening with other patches; can you please follow up on it?

> Test-patch's method of checking for new tests is not correct
> ------------------------------------------------------------
>
>                 Key: HADOOP-6164
>                 URL: https://issues.apache.org/jira/browse/HADOOP-6164
>             Project: Hadoop Common
>          Issue Type: Bug
>          Components: test
>            Reporter: Jakob Homan
>            Assignee: Giridharan Kesavan
>
> As happened in HDFS-458, test-patch/hudson saw the previous patch(HDFS-492) add a new test, bringing the total number of tests to 49.  Then it tested 458, which had no new tests as it was a change to the build file.  492 had not yet been committed yet, so there were still only 48 tests in trunk.  But test-patch failed the patch, expecting 49 tests.  test-patch should check the correct number of tests based on trunk, not on whatever number it last saw.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Commented: (HADOOP-6164) Test-patch's method of checking for new tests is not correct

Posted by "gary murry (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/HADOOP-6164?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12734805#action_12734805 ] 

gary murry commented on HADOOP-6164:
------------------------------------

Hi Giri,

If another jira is being opened for this, can we make sure an link the issues for tracking?  I think it is agreed that this is not a test-patch issue, but rather a hudson issue.

Thanks,
-Gary

> Test-patch's method of checking for new tests is not correct
> ------------------------------------------------------------
>
>                 Key: HADOOP-6164
>                 URL: https://issues.apache.org/jira/browse/HADOOP-6164
>             Project: Hadoop Common
>          Issue Type: Bug
>          Components: test
>            Reporter: Jakob Homan
>            Assignee: Giridharan Kesavan
>
> As happened in HDFS-458, test-patch/hudson saw the previous patch(HDFS-492) add a new test, bringing the total number of tests to 49.  Then it tested 458, which had no new tests as it was a change to the build file.  492 had not yet been committed yet, so there were still only 48 tests in trunk.  But test-patch failed the patch, expecting 49 tests.  test-patch should check the correct number of tests based on trunk, not on whatever number it last saw.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Updated: (HADOOP-6164) Test-patch's method of checking for new tests is not correct

Posted by "Jakob Homan (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/HADOOP-6164?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Jakob Homan updated HADOOP-6164:
--------------------------------

    Description: As happened in HDFS-458, test-patch/hudson saw the previous patch(HDFS-492) add a new test, bringing the total number of tests to 49.  Then it tested 458, which had no new tests as it was a change to the build file.  492 had not yet been committed yet, so there were still only 48 tests in trunk.  But test-patch failed the patch, expecting 49 tests.  test-patch should check the correct number of tests based on trunk, not on whatever number it last saw.  (was: As happened in HDFS-458, test-patch/hudson saw the previous patch(HDFS-492) add a new test, bringing the total number of tests to 48.  Then it tested 458, which had no new tests as it was a change to the build file.  492 had not yet been committed yet, so there were still only 48 tests in trunk.  But test-patch failed the patch, expecting 49 tests.  test-patch should check the correct number of tests based on trunk, not on whatever number it last saw.)

Fixed error in description in re: # of tests.  Otherwise issue made no sense.

> Test-patch's method of checking for new tests is not correct
> ------------------------------------------------------------
>
>                 Key: HADOOP-6164
>                 URL: https://issues.apache.org/jira/browse/HADOOP-6164
>             Project: Hadoop Common
>          Issue Type: Bug
>          Components: test
>            Reporter: Jakob Homan
>
> As happened in HDFS-458, test-patch/hudson saw the previous patch(HDFS-492) add a new test, bringing the total number of tests to 49.  Then it tested 458, which had no new tests as it was a change to the build file.  492 had not yet been committed yet, so there were still only 48 tests in trunk.  But test-patch failed the patch, expecting 49 tests.  test-patch should check the correct number of tests based on trunk, not on whatever number it last saw.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Commented: (HADOOP-6164) Test-patch's method of checking for new tests is not correct

Posted by "Giridharan Kesavan (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/HADOOP-6164?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12734456#action_12734456 ] 

Giridharan Kesavan commented on HADOOP-6164:
--------------------------------------------

my bad, I mistook hdfs-458 for hdfs-450 and made the inference.

But looking the core-test console log:
http://hudson.zones.apache.org/hudson/job/Hdfs-Patch-vesta.apache.org/25/console

     [exec]     [junit] Tests run: 1, Failures: 1, Errors: 0, Time elapsed: 3.745 sec
     [exec]     [junit] Test org.apache.hadoop.hdfs.server.datanode.TestInterDatanodeProtocol FAILED

TestInterDatanodeProtocol test failed for some reason; and hudson while parsing the xml results didn't consider the test failure.

since there is a failure in the test-core ant target test-patch script gave a -1 for this patch for core test.

I'm not sure why hudson didn't consider this failure while generating the test report.
I closed this invalid b'coz I pretty much know test-patch doesn't consider the no of tests being run by previous and current hudson build jobs.

http://hudson.zones.apache.org/hudson/job/Hdfs-Patch-vesta.apache.org/25/testReport/org.apache.hadoop.hdfs.server.datanode/TestInterDatanodeProtocol/

If you think I'm missing something please reopen this jira.

Thanks,

> Test-patch's method of checking for new tests is not correct
> ------------------------------------------------------------
>
>                 Key: HADOOP-6164
>                 URL: https://issues.apache.org/jira/browse/HADOOP-6164
>             Project: Hadoop Common
>          Issue Type: Bug
>          Components: test
>            Reporter: Jakob Homan
>            Assignee: Giridharan Kesavan
>
> As happened in HDFS-458, test-patch/hudson saw the previous patch(HDFS-492) add a new test, bringing the total number of tests to 49.  Then it tested 458, which had no new tests as it was a change to the build file.  492 had not yet been committed yet, so there were still only 48 tests in trunk.  But test-patch failed the patch, expecting 49 tests.  test-patch should check the correct number of tests based on trunk, not on whatever number it last saw.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.