-
Bug
-
Resolution: Duplicate
-
Critical
-
None
-
Platform: All, OS: All
Hudson #1.206
- One master - 0 executors
- 4 slaves (lin1, lin2, sol1, xp1) - each has one executor
- *NIX slaves are connected via ssh, XP slave via JNLP
- created on master matrix job, using all slaves, that invokes some ant script
- no jobs are running, all machines are idle
- when i schedule this job, situation is following:
on machines are deployed jobs, that are running, BUT:
1) on lin1 slave is deployed job without label (e.g. Jobname #1), never ending
2) job with label for lin1 slave is in build queue ( Jobname #1 >> lin1)
kind of deadlock, since job with label is in queue waiting for executor, but
executor is occupied with unlabled job (this unlabeled job has in console output
that it's waiting for free executor, and does not the real work - ant script)
In general, five jobs are generated - four with labels (slave names) and one
without name. The one without name occupies some machine and therefor blocks the
execution on this machine.
Feel free to ask for logs, screenshots, further information. Thanks.
- duplicates
-
JENKINS-936 Matrix parent build shouldn't consume an executor.
- Closed