Stoppable async Apex: Batch finish method vs. Queueable cancellation checks

Short answer

As of the current Apex Developer Guide, there is no documented way to interrupt an execute() or executeBatch() transaction once it has started on the server. Cancellation is only effective before the transaction dequeues or, for chained Queueable jobs, between links.

Queueable Apex

The only supported pattern to stop further work in a Queueable chain is to check a custom cancellation flag at the start of each execute method and, when the flag is set, return without enqueuing the next job.

1. Define a static Boolean (or a custom setting) that external code can flip to signal cancellation.
2. At the top of each Queueable’s execute, read the flag.
3. If the flag is true, log a cancellation event and do not chain another Queueable (return or null).
4. If the flag is false, perform the unit of work and, if needed, chain the next Queueable.

This approach prevents remaining links from executing; there is no system‑level abort for an in‑flight Queueable transaction.

Batch Apex

Batch jobs expose a finish method that runs after all scopes complete normally. If you call System.abortJob(jobId) while a batch is running:

• The abort request is best‑effort; the currently executing scope may finish before the job is marked as Aborted.
• Whether finish runs depends on the timing of the abort relative to the batch’s internal state. In many releases finish is skipped when abortion occurs after any scope has started.
• Concurrent batch limits can block rescheduling of a new batch after an abort, so you cannot rely on being able to retry immediately.

Therefore, treat finish as a guaranteed cleanup hook only for batches that complete without being aborted. For abort scenarios, place any essential cleanup logic inside the scope’s execute method (e.g., rollback‑style checks) or verify the job’s status after calling System.abortJob and act accordingly.

Verification steps (if you need to confirm behavior in your org)

1. Schedule a test Batch that sets a flag in finish and logs the start of each scope.
2. From anonymous Apex or another thread, obtain the job’s AsyncApexJob.Id and call System.abortJob(jobId).
3. Inspect Debug Logs: check whether the finish method’s log appears and review the final Status of the AsyncApexJob (should be Aborted or Completed depending on timing).
4. Repeat the same test with a Queueable chain that checks a cancellation flag after each link to confirm that subsequent links are skipped when the flag is set.

Missing diagnostic detail

If your org is near the concurrent batch limit, knowing the current limit and usage would determine whether you can safely reschedule a new batch after an abort. Please confirm your org’s API version and the value returned by Limits.getLimitBatchJobs() versus Limits.getBatchJobs() to refine the recommendation.