-
Bug
-
Resolution: Fixed
-
Critical
-
None
-
Windows
I have a top level job X. In X, I am using kicking off job Y using the File Parameter Factory option in parameterized builds. The configuration files contain a property, axis_values, which is a space separated list of different axis's for job Y.
Let's say that job X has one configuration file and it contains axis_values=A B C. I kick off job X, which in turn kicks one instance of job Y. Job Y iterates correctly through A, B and C.
Next, I change the configuration file and I set axis_values=A B C D E. I kick off job X and job Y starts. Job Y runs correctly for A, B and C. But, job Y fails to run D and E. It says that "D appears to be cancelled" and "E appears to be cancelled".
I see the following in the log:
31-May-2013 3:57:48 PM hudson.model.Executor run
SEVERE: Executor threw an exception
java.lang.NullPointerException
at hudson.model.Executor.run(Executor.java:238)
The only workaround that I have found is that it will work correctly if I put the new values at the start of the axis. (ie. axis_values=D E A B C)
List of (possibly) relevant plugins installed;
- dynamic axis
- environment injector
- jenkins parameterized trigger
- matrix sorter
- node and label parameter plugin
- is related to
-
JENKINS-24110 NPE in hudson.model.Executor when a task creates null Executable
-
- Resolved
-
-
-
- Resolved
-
[JENKINS-18164] Matrix Jobs Do Not Get Run When Updating a Dynamic Axis
tim johnston
created issue -
Greg MacDonald
added a comment - The only way I could make the problem go away was to add another row to my matrix instead of having just one row. I tried adding extra parameters but it didn't work. Seems it doesn't like single row matrices.
Greg MacDonald
added a comment - The only way I could make the problem go away was to add another row to my matrix instead of having just one row. I tried adding extra parameters but it didn't work. Seems it doesn't like single row matrices. 
tim johnston
added a comment - The problem still exists in 1.520, although the line of the exception has changed:
27-Jun-2013 10:43:45 AM hudson.model.Executor run
SEVERE: Executor threw an exception
java.lang.NullPointerException
at hudson.model.Executor.run(Executor.java:244)
tim johnston
added a comment - The problem still exists in 1.520, although the line of the exception has changed:
27-Jun-2013 10:43:45 AM hudson.model.Executor run
SEVERE: Executor threw an exception
java.lang.NullPointerException
at hudson.model.Executor.run(Executor.java:244) tim johnston
added a comment - Some more log output. It seems like the workUnit is fine, because the two pieces of axis data displayed in hudson.model.queue.WorkUnit@e617 is correct. For some reason, it appears that task.createExecutable() is returning null.
FINE: Executor #0 for QMWFCRP1 completed null in 0ms
Jul 23, 2013 6:31:04 PM hudson.model.Executor run
SEVERE: Executor threw an exception
java.lang.NullPointerException
at hudson.model.Executor.run(Executor.java:234)
Jul 23, 2013 6:31:04 PM hudson.model.Executor
FINE: Executor #0 for QMWFCRP1 is going to execute null
Jul 23, 2013 6:31:04 PM hudson.model.Executor
FINE: Executor #0 for QMWFCRP1 grabbed hudson.model.queue.WorkUnit@e617[work=RunJTF » WFS_Hours_Common_SCTAndCT,All.WFS.MSS.07.00.01.parfixes] from queue
tim johnston
added a comment - Some more log output. It seems like the workUnit is fine, because the two pieces of axis data displayed in hudson.model.queue.WorkUnit@e617 is correct. For some reason, it appears that task.createExecutable() is returning null.
FINE: Executor #0 for QMWFCRP1 completed null in 0ms
Jul 23, 2013 6:31:04 PM hudson.model.Executor run
SEVERE: Executor threw an exception
java.lang.NullPointerException
at hudson.model.Executor.run(Executor.java:234)
Jul 23, 2013 6:31:04 PM hudson.model.Executor
FINE: Executor #0 for QMWFCRP1 is going to execute null
Jul 23, 2013 6:31:04 PM hudson.model.Executor
FINE: Executor #0 for QMWFCRP1 grabbed hudson.model.queue.WorkUnit@e617 [work=RunJTF » WFS_Hours_Common_SCTAndCT,All.WFS.MSS.07.00.01.parfixes] from queue 
Brian Murrell
added a comment - Ah ha! I think this is what I am running into also. I posted about it on the mailing list.
I am running 1.534 (previously 1.530 which also exhibited the same behaviour). My matrix consists of 2 axis and results in a matrix of 1x2. It might seem silly that we have an axis of 1 item but it's there for both historical reasons (to maintain our current URLs with the same matrix format as they had historically when the first axis had more than 1 item) and future possibilities on that axis.
I did try adding a second item to the first axis to make it a 2x2 matrix but that didn't seem to resolve the issue. I have yet to find a workaround that does in my situation.
Brian Murrell
added a comment - Ah ha! I think this is what I am running into also. I posted about it on the mailing list.
I am running 1.534 (previously 1.530 which also exhibited the same behaviour). My matrix consists of 2 axis and results in a matrix of 1x2. It might seem silly that we have an axis of 1 item but it's there for both historical reasons (to maintain our current URLs with the same matrix format as they had historically when the first axis had more than 1 item) and future possibilities on that axis.
I did try adding a second item to the first axis to make it a 2x2 matrix but that didn't seem to resolve the issue. I have yet to find a workaround that does in my situation. Brian Murrell
added a comment - Just tried this on the latest LTS 1.509.3 and it has the same problem, sadly.
Brian Murrell
added a comment - Just tried this on the latest LTS 1.509.3 and it has the same problem, sadly. 
Lev Tartakovsky
added a comment - We've encountered similar problem after Jenkins crash.
We're using multi-dimension matrix jobs, and one of our matrix configurations stopped working.
I've found that affected job's matrix configuration persisting in <JENKINS HOME>/jobs/<Matrix JOB>/configurations/axis-<A>/<Value A>/axis-<B>/<Value B>/config.xml "disabled" element value left with value true.
<disabled>true</disabled>
After changing it's value to false, and reloading Jenkins configuration (Reload Configuration from Disk or restart Jenkins) my matrix job has started working.
Lev Tartakovsky
added a comment - We've encountered similar problem after Jenkins crash.
We're using multi-dimension matrix jobs, and one of our matrix configurations stopped working.
I've found that affected job's matrix configuration persisting in <JENKINS HOME>/jobs/<Matrix JOB>/configurations/axis-<A>/<Value A>/axis-<B>/<Value B>/config.xml "disabled" element value left with value true.
<disabled>true</disabled>
After changing it's value to false, and reloading Jenkins configuration (Reload Configuration from Disk or restart Jenkins) my matrix job has started working.
| Assignee | New: Oleg Nenashev [ oleg_nenashev ] |
| Link |
New:
This issue is related to |
I am having this same issue with the same null exception. Only I set my axis_values once; it ran fine for a while then started exhibiting this behavior - maybe after an update, I can't remember. My matrix build only has one row, no slaves, and everything is set to run non-concurrently. My other matrix build which is 2x2 doesn't seem to be affected.