We have been seeing a consistent "Slave went offline" message when using the docker plugin to schedule our ssh build slaves. The build appears to be working fine. I can attach to the slave container and still see the slave.jar process as well as the build job running inside the container.

However, when I look at the Nodes page, it looks like eventually the master loses connection with the slave and it starts counting down to failure. The build eventually fails after 15-17 mins. I have disabled the slave and master ping threads documented https://wiki.jenkins-ci.org/display/JENKINS/Ping+Thread

We started experiencing this issue after jumping from docker-plugin 0.9.* to 0.16.*