Try jobs and swarming tasks can time out on the pipeline side (pipeline waits too long for the task to complete and gives up), but metrics are not in place yet for measuring this.