Sometimes we experience the following problem: the job, configured to use the remove trigger plugin, is polling the remove job correctly, then a network problem occurs and the polling stops (at least judging from the logs) and after that, the job cannot be aborted, neither by plugin, nor by hand.

What I've found, the the code sets the connection timeout but it does not set the read timeout, which is by default -1/0 meaning infinite.

Log:

...
04:44:38 Waiting for remote build to finish.
04:44:38 Waiting for 30 seconds until next poll.
04:45:09 Waiting for remote build to finish.
04:45:09 Waiting for 30 seconds until next poll.
06:38:26 Build timed out (after 120 minutes). Marking the build as aborted.
...