Create Watch

A Watch in Automation Watchdog is the central mechanism configured to monitor workflows by expecting Check-ins at regular intervals according to a predefined schedule or upon activation, transitioning between Relax and Error states based on whether Check-ins occur as expected​.

Two types of Watches are:

• Scheduled Watch: Actively monitors workflows during predefined times, ensuring that expected Check-ins occur within the scheduled period. Alerts are generated if Check-ins are missed during these active windows.

• Triggered Watch: Offers dynamic activation and deactivation based on specific external events, such as when new tasks are added to a work queue or when all tasks in a queue are completed, ensuring real-time adaptability to workflow changes.

Types of Watch Periods

Once a Watch is active, it expects Check-ins based on one of the following two period types:

1. Fixed Period:

• A Check-in is required within each predefined period.
• The next period begins immediately after the previous one ends.
• If a Check-in does not occur within the period, the Watch enters an Error state.

2. Cascading Period:

• The timing of the next period is determined by when the last Check-in occurred.
• When a Check-in is received, the next period's end is set based on the defined period length.
• This approach dynamically adjusts based on workflow behavior.

Example:

• Fixed Period: If the period is 30 minutes, Check-ins are expected at 9:00 AM, 9:30 AM, 10:00 AM, and so on.

• Cascading Period: If the period is 30 minutes and the last Check-in occurred at 9:15 AM, the next Check-in is expected by 9:45 AM.

Steps to Create a Watch

1. Navigate to the Watches Page From the Watches page, click Create to initiate the setup process.

2. Provide Basic Information Complete the following fields:

   • Name: Enter a meaningful name for the Watch.
   • Description: Provide a brief description of the process the Watch monitors and its purpose. Note a single worklow may have multiple watches.

3. Select the Watch Type Choose between:

   • Missed Action: Monitors for missed check-ins and triggers alerts if none are received within the defined period.
   • Notify Only: Sends notifications whenever a check-in is received but does not monitor for missed check-ins.

   Configure a Notify Only Watch If you selected Notify Only, you can optionally configure:

   • Alert Message: Enter custom text to include in the outbound notification email. This message will be displayed in the email sent to your team whenever a check-in is received. Use this to provide context about what the check-in represents (e.g., "Daily report generation completed successfully").

4. Configure a Missed Action Watch If you selected Missed Action, complete the following additional settings:

   1. Check-in Period Type: Specify how the next required check-in time is calculated: Choose between:

      • Cascading: The next required check-in time is determined relative to the time the last check-in was received. For example, a Cascading Watch with a Period of 30 minutes: if a check-in is received at 08:48, the next check-in is due by 09:18 (30 minutes after the last check-in).
      • Fixed: The next check-in is expected at a fixed interval, regardless of when the last check-in occurred. For example, a Fixed Watch with a Period of 30 minutes, if a check-in is received at 08:48, the next check-in is due by 09:30 (the end of next 30 minute period).

   2. Check-in Period: Define the length of time until the next check-in is expected (e.g., 30 minutes).

   3. Schedule: (Optional) Define the days and times the Watch is active. Alerts are only sent during active periods. This is much like setting a repeating calendar invite. If a schedule is not used, the watch must be activated and deactivated through a separate API call to the watch or manually through the web application. For Example, a Dispatcher successfully loads work to a queue, the Dispatcher calls the API and activates the Performer Watch.

      If using a Schedule, enter the Schedule details:

      • Time Zone: Select the time zone the schedule should adhere to.
      • Start Date / Start Time: Select the date and time the schedule should commence.
      • Schedule Frequencey / Interval: Select the type of frequency of the schedule and its interval. An Interval of 1 will occur for every iteration of the frequency. Any number greater than one will occur at each interval, e.g. Weekly with Interval of 4, results with the Watch being active once every 4 weeks.
      • Days of the Week: Select days of the week if you want to restrict the schedule to specific days.
      • By Hour: To restrict the schedule to specific hours of the day, enter them as a comma separated list.
      • By Minute: To restrict the schedule to specific minutes (e.g. 0, 30), enter them as a comma separated list.
      • Duration: Set the duration to how many mintues the watch should be active on the repeating schedule. At the end of the duration, the Watch will automatically become inactive. For Example, to have the watch active for 3 hours, set Duration to 180 minutes.

   4. Activate Watch: (Optional) Select another watch to activate when this watch deactivates. This supports handoff patterns where one workflow finishes and another should immediately begin monitoring. For Example, a dispatcher watch can activate a performer watch when the dispatcher deactivates after loading work. The selected target watch must not have a schedule. See Activate On Deactivation for more details.

   5. Condition By Queue: (Optional) Allows you to configure a watch for the unique case that a workflow may work against different queues. Provide the names of the queues that should be monitored. For Example, the same workflow runs against different queues, one queue for each of your customers, this will allow you to monitor each of the queues independently without having to create multiple watches.

   6. Condition By Machine: (Optional) Enable this setting when each machine running the workflow should be monitored independently. When enabled, each machine that checks in will have its own watch condition with independent state, check-in deadlines, and threshold tracking. Machines register automatically on their first check-in, and on non-scheduled watches they can also activate the watch in that same request when Activate On Check-In is enabled. For Example, a workflow runs on 5 machines and each machine must independently check in every 15 minutes. With Condition By Machine enabled, if Machine 3 misses a check-in, only Machine 3 enters an error state while the others remain in relax. See Condition By Machine for more details.

   7. Expected Machine Count: (Optional) Available for Standard and Condition By Machine watches. Use this when a run is expected to involve more than one machine, and the watch should wait for a minimum number of machines to report before it clears its overdue state. Leave it empty to disable. For Example, if a job should fan out to 3 workers, set Expected Machine Count to 3 so the watch does not return to relax until all 3 machines have reported. See Expected Machine Count for more details.

   8. Activate On Check-In: (Optional) Enable this setting to allow the watch to automatically activate when it receives its first check-in. This is only available for watches without a schedule. It can be used with Standard, Condition By Queue, and Condition By Machine watches. When combined with Condition By Machine, the first machine check-in can register the machine and activate the watch in a single request. See Activate On Check-In for more details.

   9. Threshold Monitoring: (Optional) Enable this setting to monitor the quality of your workflow's check-ins. When enabled, configure the following:

      • Window Size: The number of recent check-in outcomes to evaluate (up to 1000).
      • Minimum Observations: The number of outcomes required before threshold evaluation begins.
      • Threshold Percent: The minimum success rate percentage required. For example, set to 80 to require at least 80% of outcomes to be successes.
      • Consecutive Failure Limit: The maximum number of consecutive failures allowed before the watch enters an error state.

      See Threshold Monitoring for more details on how outcomes and thresholds work.

5. Save Review the configured settings and click Create. If a Schedule was used, the Watch will by default be Active and in a Relax State. If no Schedule was set, the Watch will by default be Inactive and in a Relax State.

   The Watch Page shall be displayed.

6. Note the Watch Id: From the Watch Page, the Watch Id is displayed in the Watch Properties section. Each Watch has a unique Id. This Id is required by the Robot for any API call for this Watch.