Clean up nodes when connection to jenkins master unexpectedly lost

[rouke-broersma]

Issue #68 could occur when jenkins master is unexpectedly shut down (kill command, server crash etc), we should have the still online slaves clean up all selenium processes when this happend.

[zdmeyers]

Just to keep you in the loop, I just tried to update to the new LTS (1.651.3) and the bug returned. Duplicate nodes on my windows slaves. However, the problem did not manifest itself on my Windows 10 VM, just Win8.1 and Win7. So, I downgraded back to 1.650, cleaned up the duplicates and restarted jenkins and no bug. I used the /exit URL to shutdown jenkins when I upgraded the version. No idea if that is significant.

[rouke-broersma]

Did the update process perhaps return the plugin back to the normal install, instead of the fixed version I sent you?

[zdmeyers]

Doesn't appear to have. I still have the private build you sent me installed according to the list of installed plugins. All I did was replace the jenkins.war file with the LTS version. I did not update plugins after that as I had already done that before upgrading and the Selenium plugin stayed the same.

[rouke-broersma]

That is.. Strange. I guess I'll take a look when I have time. Currently I'm busy finishing my internship.