Skip to content
Dan Mushkevych edited this page Dec 9, 2016 · 3 revisions

Garbage Collector's algorithm: Selects all units of work satisfying next criteria and re-triggers them:

  • Unit of work must have created_at timestamp within the last 48 hours
  • Unit of work must be in STATE_IN_PROGRESS, STATE_REQUESTED, STATE_INVALID states
  • Unit of work must belong to a process that is both registered in the ProcessContext and marked as active
  • Unit of work must not belong to free-run process

Selected unit_of_work are placed into a queue, where they spend 15 minutes, unless manually re-triggering from MX UI. During re-triggering, Garbage Collector performs following actions:

  • Changes unit_of_work state to STATE_REQUESTED
  • Increments number_of_retries by one
  • Fetches proper MQ Publisher and re-sends the unit_of_work

Note: in case unit_of_work has crossed 48 hour-life-support time frame, Garbage Collector will change its state to STATE_CANCELED. As soon as the Synergy Scheduler sees a unit_of_work in state STATE_CANCELED, it marks the timeperiod as STATE_SKIPPED.