When triggering a remote build the plugin tries to determine the next job number in advance. After calling start on the remote job It then checks if the job is started.

In our case the first call to our job returns an empty response, probably because it takes too long to initialize. The plugin then tries to iterate older jobs and probably even finds one that has the same paramters. This can then lead to not waiting for the job to be build.

In our case this is critical because we are starting several remote jobs that need to be run sequentially. So the next job gets startet too early and both jobs are run in parallel when they should be executed sequentially.

Workaround: Add a timestamp parameter to the build so that all execution parameter sets are unique.

Pull request with a better fix coming soon.,