-
Bug
-
Resolution: Unresolved
-
Critical
-
Window 10
Jenkins 2.51
The pipeline batch command failed 3 out of 4 times and hang mostly after a long command. Both master and slave node are waiting for each other. Not sure it's the same, but here's what I have:
- Jenkins 2.51
- Windows 10 slave
- Linux Master (CentOS 7)
- pipeline script from SCM
- Build Trigger is Poll SCM (manual trigger build does not have this behavior and complete successfully)
- Mercurial SCM
- The session is lock during the job is executing (user is still logon and slave is still available)
- Seem to always happen on long batch command (short one doesn't display this behavior or maybe it's just less likely)
- The project is parametrized for pipeline script repos and revision (default value are provided and the proper checkout is made).
- Seem like the command complete successfully I see the final data into the log but it look like the master/slave doesn't known the batch command have terminated
- I use the following syntax:
bat returnStatus: false, script: 'msbuild ...'
I cannot stop/cancel the build. I have to restart the master to unjam the slave and master (killing the slave client doesn't do anything either).
Here's the last things into the console log:
18:00:58 18:00:58 Build succeeded. 18:00:58 0 Warning(s) 18:00:58 0 Error(s) 18:00:58 18:00:58 Time Elapsed 00:15:41.55
which is correct, indicate to me that the msbuild command finished properly.
This is a total show stopper, we cannot have any more CI with this behavior, we always have to restart the master. Make us wonder if we should start looking for an alternative (I have report this issue into the forum thread, without any answer 3 times already). The batch command seem to hang for many people if I see the bug listing, we all have different system and setup, but they all are related to the batch command seem like a nightmare for hang. Some are marked as resolved and many are still open.