Skip to content
Epiglossary

Schedule a function

Kinetic + Classic

Any function can be run by the System Agent on a schedule: nightly data housekeeping, an hourly sync with another system, a weekly summary email. This replaces the old habit of hanging logic on a directive that “happens to fire often enough”.

Before you start

Section titled “Before you start”
  • The function’s library must be authorized for the company you schedule it from (its Security card). A scheduled call from an unauthorized company fails.
  • Decide what “success” looks like. A scheduled function runs asynchronously, so nobody receives its response parameters. If you need to know what it did, the function has to tell you (see below).

Steps

Section titled “Steps”
  1. Create a schedule in System Agent Maintenance (System Setup > System Maintenance > System Agent Maintenance): choose New Schedule, give it a description, set the Schedule Type (for example Interval, for “every N minutes”) and its timing, and save. Skip this if a suitable schedule already exists.
  2. Open Schedule Epicor Function (under System Management > Business Process Management in the Classic menu).
  3. Select the Library ID and the Function ID.
  4. Enter values for the function’s request parameters.
  5. Pick the Schedule and tick Recurring.
  6. Click Submit.

Checking that it ran

Section titled “Checking that it ran”

Each run appears in System Monitor as a Run Epicor Function task. History Tasks shows completed and failed runs; Scheduled Tasks shows what’s queued. To stop a recurring schedule, select it on Scheduled Tasks and delete it.

Designing a function for the scheduler

Section titled “Designing a function for the scheduler”

A scheduled function has no user watching it. Build that in:

  • Report the outcome yourself. Email a short summary (“42 customers checked, 3 updated”), or write a row to a UD table that a dashboard shows. Info messages go nowhere.
  • Send one summary, not one email per problem. A job that fails for 2,000 records shouldn’t send 2,000 emails. Collect the failures and send one message with a clear subject, such as [FAILURE] Nightly credit review.
  • Log detail to the server log for troubleshooting, using the application logger. See Troubleshooting functions.
  • Make it safe to run twice. Only change records that still need changing, so a re-run after a failure, or two overlapping runs, don’t do the work twice.
  • Keep each run short. If a job takes longer than its interval, runs start to overlap. Process in batches or lengthen the interval.
  • Retry transient errors. Record locks and “row has been modified by another user” conflicts often succeed on a second attempt a moment later.

Good candidates

Section titled “Good candidates”