Jenkins build fails on slaves with java.lang.InterruptedException

Question

Jenkins build fails with Cannot contact XXXXXXXXXXXX: java.lang.InterruptedException

from time to time. it doesn't matter if it is a spot instance or an on-demand instance

Jenkins ver. 2.60.3 
Amazon EC2 plugin 1.36

Answer 1

I monitored the slave stats and figured out that the slave was under heavy load like 1200. The slaves build a lot of docker images. The load was due to the fact that the save was I/O bound. The docker volume /var/lib/docker and workspace were EBS and EFS mounts respectively. Upgrading the linux kernel and changing the docker volume to Overlay2 solved the issue.

Answer 2

In my case after the error described above in the pipeline, I checked the specific node console log and it appeared the following error:

java.util.concurrent.TimeoutException: Ping started on XXXX hasn't completed at XXX

Probably the jenkins slave was corrupted. Just rebooting the slave solved the problem as it is explained in this jenkins issue