Scheduled jobs need a stable run identity, controlled execution and a way to recover partial progress. A scheduler triggering once is not proof that business work happens exactly once. Define overlap, retries, missed schedules and external side effects before relying on a background job for important operations.
Let’s plan your scheduled job recovery.
Bring your workflow, constraints and open questions. We’ll help you define the requirements and work through a practical next step.
Define the logical run
Identify the population and period a run processes. Keep that logical identity separate from technical attempts. A retry of yesterday's settlement preparation should not become a second independent run merely because it has a new process identifier.
Microsoft's Scheduler Agent Supervisor pattern discusses durable state and recovery for distributed work. Use the architecture appropriate to the task rather than assuming a recurring timer solves coordination.
Specify execution scenarios
| Scenario | Rule to agree | Evidence to retain |
|---|---|---|
| Previous run still active | Skip, queue or permitted overlap | Reason and active-run reference |
| Worker stops mid-run | Resume from a safe point | Durable item or step state |
| Schedule missed | Defined catch-up or expiry | Periods covered and omitted |
| External call times out | Verify outcome before repeating | Provider and operation reference |
| Manual rerun | Authorised scope and duplicate protection | Actor and intended population |
| Configuration changes | Apply a known version | Effective settings for the run |
Locks and leases need failure behaviour too. Define expiry, renewal and ownership so a stopped worker does not leave work permanently blocked or allow conflicting execution.
Separate completion from partial progress
For a hypothetical daily reminder job, some messages may be sent before a failure. Retrying the entire population without per-item tracking can duplicate notifications. Record which items reached the defined outcome.
Show run totals with meaningful categories: eligible, attempted, completed, skipped and unresolved. A single green status should not conceal missing work.
Exercise recovery deliberately
Stop a test worker after a controlled step, start overlapping triggers and rerun a completed period. Verify the resulting business records and external side effects, not only the job status table.
Assign ownership for aged runs and repeated failures. Include a safe operational procedure for pausing work while investigating uncertainty.
Scheduled-job questions
Does a distributed lock guarantee exactly-once outcomes?
Not across every failure boundary. Side effects and recovery need explicit duplicate protection and reconciliation.
Should a missed job always run immediately later?
Only if the work is still useful and safe. Some tasks expire or require review after their deadline.
Pair execution rules with integration monitoring, and discuss the workflow with Drema's cloud team.
This article explains what to build, not legal advice. Rules change; confirm against the current official text before relying on it.