After a job has spawned a background process on a windows slave, the job cannot be aborted or killed.
Klicking "Abort" has no effect.
Klicking "Kill" will stop the job itself, but all processes of the job keeps executing on the slave machine as if nothing happened.

The problem can be reproduced using this groovy script:

node() {
   bat "start ping localhost -n 100"
   bat "ping localhost -n 100"
}

I have tried the same on a Linux slave, and found no issues there.

The problem is repeatable on 10 out of our 15 windows slaves. Comparing remoting version, JVM version and OS version on all these slaves, I find nothing which correlates to the machines where the problem occurs.