Uploaded image for project: 'Jenkins'
  1. Jenkins
  2. JENKINS-14814

"Ping-pong" builds store excessively large CauseAction

    Details

    • Similar Issues:

      Description

      Create two jobs, a and b, each of which does nothing but trigger the other. (Useless by itself, but a similar idiom can be used to alternate indefinitely between two actions, such as checking for changes in some part of the environment and reacting to those changes.)

      Let this run for a few dozen iterations. The build.xml file for each build can soon run into multiple megabytes. For example, on just the fourth cycle, a/builds/4/build.xml has this lengthy "cause" section:

      <hudson.model.CauseAction>
        <causes>
          <hudson.model.Cause_-UpstreamCause>
            <upstreamProject>b</upstreamProject>
            <upstreamUrl>job/b/</upstreamUrl>
            <upstreamBuild>3</upstreamBuild>
            <upstreamCauses>
              <hudson.model.Cause_-UpstreamCause>
                <upstreamProject>a</upstreamProject>
                <upstreamUrl>job/a/</upstreamUrl>
                <upstreamBuild>3</upstreamBuild>
                <upstreamCauses>
                  <hudson.model.Cause_-UpstreamCause>
                    <upstreamProject>b</upstreamProject>
                    <upstreamUrl>job/b/</upstreamUrl>
                    <upstreamBuild>2</upstreamBuild>
                    <upstreamCauses>
                      <hudson.model.Cause_-UpstreamCause>
                        <upstreamProject>a</upstreamProject>
                        <upstreamUrl>job/a/</upstreamUrl>
                        <upstreamBuild>2</upstreamBuild>
                        <upstreamCauses>
                          <hudson.model.Cause_-UpstreamCause>
                            <upstreamProject>b</upstreamProject>
                            <upstreamUrl>job/b/</upstreamUrl>
                            <upstreamBuild>1</upstreamBuild>
                            <upstreamCauses>
                              <hudson.model.Cause_-UpstreamCause>
                                <upstreamProject>a</upstreamProject>
                                <upstreamUrl>job/a/</upstreamUrl>
                                <upstreamBuild>1</upstreamBuild>
                                <upstreamCauses>
                                  <hudson.model.Cause_-UserIdCause/>
                                </upstreamCauses>
                              </hudson.model.Cause_-UpstreamCause>
                            </upstreamCauses>
                          </hudson.model.Cause_-UpstreamCause>
                        </upstreamCauses>
                      </hudson.model.Cause_-UpstreamCause>
                    </upstreamCauses>
                  </hudson.model.Cause_-UpstreamCause>
                </upstreamCauses>
              </hudson.model.Cause_-UpstreamCause>
            </upstreamCauses>
          </hudson.model.Cause_-UpstreamCause>
        </causes>
      </hudson.model.CauseAction>
      

      Excessive disk usage consumption is not the only problem; Jenkins may even refuse to start, since loading one of these files can produce a stack overflow:

      hudson.util.RobustReflectionConverter.unmarshal(RobustReflectionConverter.java:180) 
      hudson.util.XStream2$PassthruConverter.unmarshal(XStream2.java:323) 
      hudson.util.XStream2$AssociatedConverterImpl.unmarshal(XStream2.java:293) 
      com.thoughtworks.xstream.core.TreeUnmarshaller.convert(TreeUnmarshaller.java:82) 
      com.thoughtworks.xstream.core.AbstractReferenceUnmarshaller.convert(AbstractReferenceUnmarshaller.java:63) 
      com.thoughtworks.xstream.core.TreeUnmarshaller.convertAnother(TreeUnmarshaller.java:76) 
      com.thoughtworks.xstream.core.TreeUnmarshaller.convertAnother(TreeUnmarshaller.java:60) 
      com.thoughtworks.xstream.converters.collections.AbstractCollectionConverter.readItem(AbstractCollectionConverter.java:71) 
      hudson.util.RobustCollectionConverter.populateCollection(RobustCollectionConverter.java:85) 
      com.thoughtworks.xstream.converters.collections.CollectionConverter.unmarshal(CollectionConverter.java:61) 
      hudson.util.RobustCollectionConverter.unmarshal(RobustCollectionConverter.java:76) 
      com.thoughtworks.xstream.core.TreeUnmarshaller.convert(TreeUnmarshaller.java:82) 
      com.thoughtworks.xstream.core.AbstractReferenceUnmarshaller.convert(AbstractReferenceUnmarshaller.java:63) 
      com.thoughtworks.xstream.core.TreeUnmarshaller.convertAnother(TreeUnmarshaller.java:76) 
      com.thoughtworks.xstream.core.TreeUnmarshaller.convertAnother(TreeUnmarshaller.java:60) 
      hudson.util.RobustReflectionConverter.unmarshalField(RobustReflectionConverter.java:290) 
      hudson.util.RobustReflectionConverter.doUnmarshal(RobustReflectionConverter.java:233) 
      hudson.util.RobustReflectionConverter.unmarshal(RobustReflectionConverter.java:180)
      

      While recording transitive causes may be useful in general, there must be some sanity cutoff point.

        Attachments

          Issue Links

            Activity

            Hide
            scm_issue_link SCM/JIRA link daemon added a comment -

            Code changed in jenkins
            User: Jesse Glick
            Path:
            changelog.html
            http://jenkins-ci.org/commit/jenkins/2aa272a0fae0390eb2da679ab7c9e0d402b1f53b
            Log:
            JENKINS-14814 Noting.

            Show
            scm_issue_link SCM/JIRA link daemon added a comment - Code changed in jenkins User: Jesse Glick Path: changelog.html http://jenkins-ci.org/commit/jenkins/2aa272a0fae0390eb2da679ab7c9e0d402b1f53b Log: JENKINS-14814 Noting.
            Hide
            jglick Jesse Glick added a comment -

            Workaround:

            perl -i.bak -n -e 'print unless m{<hudson\.model\.CauseAction>}..m{</hudson\.model\.CauseAction>}' jobs/*/builds/*-*/build.xml
            
            Show
            jglick Jesse Glick added a comment - Workaround: perl -i.bak -n -e 'print unless m{<hudson\.model\.CauseAction>}..m{</hudson\.model\.CauseAction>}' jobs/*/builds/*-*/build.xml
            Hide
            dogfood dogfood added a comment -

            Integrated in jenkins_main_trunk #1856
            JENKINS-14814 Reproduced. (Revision b0278767831d751b93ad77dc2c5c92b6d140e865)
            [FIXED JENKINS-14814] Truncate deeply nested upstream causes. (Revision bfa648152f266540536e76f7b3ebcb1598ad5f6f)
            JENKINS-14814 Noting. (Revision 2aa272a0fae0390eb2da679ab7c9e0d402b1f53b)

            Result = UNSTABLE
            Jesse Glick : b0278767831d751b93ad77dc2c5c92b6d140e865
            Files :

            • test/src/test/java/hudson/model/CauseTest.java

            Jesse Glick : bfa648152f266540536e76f7b3ebcb1598ad5f6f
            Files :

            • core/src/main/java/hudson/model/Cause.java

            Jesse Glick : 2aa272a0fae0390eb2da679ab7c9e0d402b1f53b
            Files :

            • changelog.html
            Show
            dogfood dogfood added a comment - Integrated in jenkins_main_trunk #1856 JENKINS-14814 Reproduced. (Revision b0278767831d751b93ad77dc2c5c92b6d140e865) [FIXED JENKINS-14814] Truncate deeply nested upstream causes. (Revision bfa648152f266540536e76f7b3ebcb1598ad5f6f) JENKINS-14814 Noting. (Revision 2aa272a0fae0390eb2da679ab7c9e0d402b1f53b) Result = UNSTABLE Jesse Glick : b0278767831d751b93ad77dc2c5c92b6d140e865 Files : test/src/test/java/hudson/model/CauseTest.java Jesse Glick : bfa648152f266540536e76f7b3ebcb1598ad5f6f Files : core/src/main/java/hudson/model/Cause.java Jesse Glick : 2aa272a0fae0390eb2da679ab7c9e0d402b1f53b Files : changelog.html
            Hide
            zxkane Meng Xin Zhu added a comment -

            Do you know which build starts to have this fix? Because I still see similar issue happened in LTS 1.480.2.

            Show
            zxkane Meng Xin Zhu added a comment - Do you know which build starts to have this fix? Because I still see similar issue happened in LTS 1.480.2.
            Hide
            jglick Jesse Glick added a comment -

            Should be in 1.479, but you might be hitting JENKINS-15747.

            Show
            jglick Jesse Glick added a comment - Should be in 1.479, but you might be hitting JENKINS-15747 .

              People

              • Assignee:
                jglick Jesse Glick
                Reporter:
                jglick Jesse Glick
              • Votes:
                0 Vote for this issue
                Watchers:
                4 Start watching this issue

                Dates

                • Created:
                  Updated:
                  Resolved: