Pipeline durability when a slave to master connection is interrupted

[Florian Kammermann]

I have a questions on master slave connectivity robustness.

Is there any documentation on the guarantees I have, when the master - slave connection is interrupted, when a pipeline is executed on the slave?

I just did a try with a stupid shell script:

sh 'while true; do sleep 1; echo $(date); done'

I interrupted the connection for 4sec, this was visible in the job log as:

Cannot contact external-slave: hudson.remoting.RequestAbortedException: java.nio.channels.ClosedChannelException

But didn't brake the job itself, after reconnect the job log continued on the master.

This was a very basic test.

But what happens if the pipeline execution goes back to master or another slave, when the connection interrupts?

Does Jenkins handle these cases correctly (eg. pause the pipeline, try to reconnect, if not possible to reconnect for ?sec, abort).

Are there timeouts configured on the pipeline execution, when a master - slave connection interrupts?