Problem Handling

One of the big problems with automated workflows is exception handling or handling problems that occur during that workflow. These can be caused by faulty jobs, faulty tools or problems with external services (such as networking).

SWITCH includes a number of different ways to make the problem handling easier and more reliable.

Keeping track of automated processing

The first step to solving problems is knowing that they happened. SWITCH includes very robust logging; every task performed or initiated by SWITCH is logged. Those logs can easily be sorted and searched in SWITCH, can be seen in the SWITCHclient from your own workstation and can be exported to a text or XML file.

Handling problem jobs

Once in a while, a particular job will cause problems. This could be because it wasn't the type of job expected by the tool used or even because the job was damaged at some point. Such jobs are quarantined by SWITCH in a problem jobs folder. Each flow in SWITCH has its own such problem jobs folder.

An operator can then be notified by email that this happened, but SWITCH also includes a problem jobs flow element that can be used to provide more customized error handling. This flow element can be used in any flow and it automatically receives all jobs that caused problems. As it is a regular flow element, other tools in SWITCH can then be attached to customize processing of those problem jobs.

Handling problem processes

Sometimes its not a particular job that causes a problem, but it is a certain process. The perfect example is networking where an email or FTP server might be down, or a shared network drive might no longer be reachable.

In those cases SWITCH does not move the affected jobs to the problem jobs folder, but it keeps retrying the action that failed until it is successfull. In the preferences the production manager can specify who should be notified of such events and from which moment (upload to an FTP server could for example be retried a number of times before someone is notified).

Keeping an eye out for trouble: the dashboard

SWITCH also includes a pane called the dashboard, that summarizes what is happening in SWITCH: which jobs are being processed, which jobs caused problems, which services are currently down. In one simple overview, it tells a production manager the status of its automated workflow.