Skip to content

cancellation.md

Latest commit

 

History

History
34 lines (22 loc) · 2.25 KB

cancellation.md

File metadata and controls

34 lines (22 loc) · 2.25 KB

Cancellation

Every user Func accepted by IDurableContext (StepAsync, RunInChildContextAsync, WaitForCallbackAsync, WaitForConditionAsync) receives a CancellationToken parameter. Pass it to cancellation-aware APIs inside the body so the workflow can tear down cleanly.

What the token observes

The token is a linked source combining:

  1. The CancellationToken you passed to the IDurableContext method (so the caller's cancel intent reaches the body).
  2. An SDK-owned workflow-shutdown signal that fires when the workflow is being torn down (a sibling operation suspended, a checkpoint failed, or a future parallel branch aborted).
var user = await ctx.StepAsync(
    async (_, ct) => await httpClient.GetAsync(url, ct),
    name: "fetch");

When either trigger fires, the token transitions to IsCancellationRequested = true and awaits on cancellation-aware APIs unwind via OperationCanceledException.

Semantics

  • OperationCanceledException thrown out of a step body via the token (i.e. linked.IsCancellationRequested is true) is treated as cancellation: no FAIL checkpoint is written, no retry is consulted. The exception propagates up.
  • OperationCanceledException thrown by user code for unrelated reasons (token never fired) is treated as a normal step failure: FAIL checkpoint, retry per the configured RetryStrategy.
  • The SDK's own writes (checkpoints, batcher flush, the runtime API response) never observe the workflow-shutdown signal. Successful work is never lost to teardown.

Guidance

  • Do pass ct into every cancellation-aware call inside the step body (HttpClient.SendAsync(ct), Task.Delay(ct), AWS SDK calls). This is what makes caller-cancel and shutdown-cancel actually unwind.
  • Don't branch workflow logic on IsCancellationRequested. It is a runtime concern, not a workflow concern; branching on it makes the workflow non-deterministic across replays.
  • Don't catch (OperationCanceledException) and continue. Either don't catch, or catch and rethrow.

Replay

Cached operations short-circuit before the user Func is invoked. A SUCCESS checkpoint replays its serialized result; the token is never built or observed. Replay determinism is structural — cancellation cannot affect it.