hadoop-common-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Apache Wiki <wikidi...@apache.org>
Subject [Hadoop Wiki] Update of "HudsonBuildServer" by NigelDaley
Date Fri, 25 Jul 2008 22:04:08 GMT
Dear Wiki user,

You have subscribed to a wiki page or wiki category on "Hadoop Wiki" for change notification.

The following page has been changed by NigelDaley:

The comment on the change is:
Add some info on fixing hudson when the patch queue seems to hang

           1. {{{ps -ef | grep hudson}}} again to verify the build and test processes are
        When this process dies, Hudson will fail the build.
+    1. '''Restarting a hung patch testing queue:'''
+       A number of issues can cause the [http://hudson.zones.apache.org/hudson/view/Hadoop/job/Hadoop-Patch-Admin/lastSuccessfulBuild/artifact/HADOOP_PatchQueue.html
queue of pending patches] to hang or appear hung.  Note that the queue page is updated by
the [http://hudson.zones.apache.org/hudson/view/Hadoop/job/Hadoop-Patch-Admin/ Hadoop-Patch-Admin]
build every five minutes.  Sometimes, this build can't run for a long period of time because
other build are running on Hudson.
+          1. '''There are patches in the queue but [http://hudson.zones.apache.org/hudson/view/Hadoop/job/Hadoop-Patch/
Hadoop-Patch] build is not running'''
+             Try kicking off the [http://hudson.zones.apache.org/hudson/view/Hadoop/job/Hadoop-Patch-Admin/
Hadoop-Patch-Admin] manually.  If that doesn't start the [http://hudson.zones.apache.org/hudson/view/Hadoop/job/Hadoop-Patch/
Hadoop-Patch] build, then kick off the [http://hudson.zones.apache.org/hudson/view/Hadoop/job/Hadoop-Patch/
Hadoop-Patch] build manually.  That should be all that is necessary in this scenario.
+          1. '''The [http://hudson.zones.apache.org/hudson/view/Hadoop/job/Hadoop-Patch/
Hadoop-Patch] is running well over 2 hours'''
+             Some test has likely hung.  See "Killing a hung test" below. 
     1. '''Killing a hung test:'''
        Sometimes it is desirable to just kill a hung test and allow the rest of the build
to continue. Although the build will still fail, it can sometimes provide useful information
if you can get a thread dump on the hung test and, if the build is in the first test pass,
the second pass might succeed as some tests hang only intermittently. You can tell a test
is hung by looking at the console log for the build and if the last entry is some time ago
(at least 15 minutes) it is safe to assume the test is hung.
-          1. Login to {{{hudson.zones.apache.org}}}
+          1. {{{ssh hudson.zones.apache.org}}}
-          1. Switch to the {{{hudson}}} user
+          1. {{{sudo su - hudson}}}
           1. Find the hung test: {{{ps -laxww | grep hudson | grep test.build.data | grep
-v grep'}}}
              For example:

View raw message