We’ve been working to implement a new timeout that will prevent Selenium failures from consuming many of our customers' minutes. This new timeout will detect issues in Selenium’s end (e.g. browser or Selenium crashing before answering back) and will stop and fail the job when that happens.
(Notice: this timeout is only triggered when Selenium doesn't send a reply back to your end after a command has been requested for a certain amount of time. Without this feature, we’ll wait for 30 minutes, or until the full job exceeds the maximum duration, before stopping the whole job).
While this feature has not yet been made publicly available, you can start using it now by setting the command-timeout capability to any value you believe reasonable.
Starting next week, and as our docs state, we'll be setting this to a default value of 300 seconds for all jobs that don't explicitly specify a timeout limit . Once this happens, you may find that some of your tests get caught and fail. We believe the default value - five minutes - is long enough so that in those special cases, you’ll be able to revisit the cause of such a long command and either fix the problem or extend the timeout.
Please let us know if you have any questions or concerns about this.