Jenkins job started by timer at an unexpected time

[ludovic.v...@atos.net]

Dear,

We are currently using Jenkins 1.596.3 and we have faced an issue with a job which was not started at the correct timestamp by the timer.

The job is configured to be executed automatically on a daily basis at 7:30pm CEST.

The job execution has been correctly triggered and the build has started at 7:30pm CEST, as expected.

However, the logs of the build show the following error message:

Looks like the node went offline during the build. Check the slave log for the details. 
FATAL: channel is already closed

It seems that there has been a communication issue between our Jenkins master (hosted on a VM) and our Jenkins slave (a physical server)

Therefore, the job has failed.

At that time, we did not face any hardware issue on the servers.

In the jenkins logs, except if I missed it, I did not find any reference of this communication issue, therefore, I cannot identify the root cause of this communication issue.

Could you please tell if there is a possibility to get more information ?

Later during the night (at 00:15am CEST), we have done a restart of the Jenkins instance and the job has been executed at 00:30am CEST by the trigger.
This execution was not expected because the job is not configured to be executed at 00:30am.

Could you please if this execution is a consequence of the communication issue faced before ?

Could you please describe the process when jobs are aborted for such issues ? Are they automatically restarted by Jenkins after its next restart ?

FYI, this request relates to https://issues.jenkins-ci.org/browse/JENKINS-31230 which has been rejected for the following reason:

Daniel Beck added a comment - 17/Nov/15 12:54 PM

This is an issue tracker, not a support site. Furthermore, 1.596.x and 1.609.x are no longer supported, you need to show a bug exists on 1.625.x to get attention in this tracker.

To get help, ask the jenkinsci-users mailing list, or in #jenkins on Freenode.

Best regards

[Victor Martinez]

Hi,

If you connect those slaves via ssh then can you review your networking configuration and verify no interruptions happened when the build ran?

You can add some verbose output based on the logs:

- https://wiki.jenkins-ci.org/display/JENKINS/Logging

SSH slave/master connections within Jenkins retries up to 5 times to get connected afaik.

Cheers

[ludovic.v...@atos.net]

Dear,

As the issue occured, we did not see any error message in the logs of https://[OUR_JENKINS_INSTANCE]/computer/[OUR_SLAVE_SERVER]/log

On which class would you advice us to add logging ?

What would be the link between the fact that a connection issue has perhaps occured between the master and the slave, and the fact that the job has been triggered at an unexpected time ?

Best regards