Which Talos API version streams enforce cancellation semantics for timed-out configuration requests?

Answer to the core question

Talos does not expose a declarative REST API; it uses a gRPC service (machined/apid) for all node‑management operations. Because cancellation is a property of the gRPC runtime, all Talos API streams enforce the same cancellation semantics regardless of the release stream. The client‑side deadline set by talosctl --timeout maps directly to a gRPC context deadline exceeded error. When this deadline expires, the client cancels the RPC and the server receives a cancellation signal. The node continues to apply the desired configuration that was already accepted; the cancellation does not roll back or abort the in‑flight apply.

Confirmed facts

• Talos nodes run the machined service, which implements a gRPC API, not a REST endpoint.
• Cancellation semantics are derived from gRPC and are consistent across all Talos releases; there are no separate “API version streams” that change this behavior.
• When a client times out, talosctl exits with a DeadlineExceeded error and a non‑zero exit code, but the node’s configuration state may still be progressing toward the last accepted spec.
• Downstream integrations can rely on idempotent apply/get calls or the machine status query to determine final state, without implementing custom polling for cancellation.

Practical steps for your environment

1. Run a quick test: talosctl --timeout 1s apply-config --config my-config.yaml against a test node.
2. Immediately check the node’s state: talosctl get machineconfig --node 10.0.0.5. The configuration should reflect the applied spec even though the client timed out.
3. Verify the exit code: echo $? should return a non‑zero value (typically 1). The error message will contain DeadlineExceeded.
4. If you need to confirm that the node has finished reconciling, run talosctl get machine --node 10.0.0.5 and look for a State: Ready or State: Updating transition.

What to check in your release notes

• Look for any mention of “context cancellation” or “DeadlineExceeded” handling in the machined source for your specific Talos version.
• Verify that the apply-config RPC handler does not ignore the context cancellation; older pre‑v1.10 releases had a known issue where the handler would continue despite a cancelled context.

Follow‑up diagnostic detail

To give you the most accurate guidance, could you let me know the talosctl and Talos node version numbers you are using? Different minor releases may have subtle fixes that affect how cancellation is logged or reported.