fix(sdks/go): recover durable waits after reconnect

[igor-kupczynski]

Description

Go durable waits now report pending callbacks after reconnect and periodically so persisted completions can be replayed.

Why:

• Stream interruptions previously lost callback routing, leaving completed durable waits stuck.
• Periodic status reconciliation allows the engine to rediscover and redeliver completed entries.

How:

• Track outstanding durable callback waits and report them through WorkerStatus after connecting.
• Resend current waiting entries whenever callback state changes and every second.
• Preserve pending callbacks across reconnects and reconcile them with engine responses.
• Atomically deliver or briefly buffer early completions, with race-safe shutdown handling.

similarity to other SDKs

Python

• hatchet/sdks/python/hatchet_sdk/clients/listeners/durable_event_listener.py

  Lines 293 to 316 in 522678b

  	async def _poll_worker_status(self) -> None:
  	if self._request_queue is None or self._worker_id is None:
  	return
  	if not self._pending_callbacks:
  	return
  	waiting = [
  	DurableTaskAwaitedCompletedEntry(
  	durable_task_external_id=task_ext_id,
  	invocation_count=inv_count,
  	node_id=node_id,
  	branch_id=branch_id,
  	)
  	for (task_ext_id, inv_count, branch_id, node_id) in self._pending_callbacks
  	]
  	request = DurableTaskRequest(
  	worker_status=DurableTaskWorkerStatusRequest(
  	worker_id=self._worker_id,
  	waiting_entries=waiting,
  	)
  	)
  	await self._request_queue.put(request)

TypeScript

• hatchet/sdks/typescript/src/clients/listeners/durable-listener/durable-listener-client.ts

  Lines 384 to 404 in 522678b

  	private _pollWorkerStatus(): void {
  	if (!this._workerId || this._pendingCallbacks.size === 0) return;
  	const waitingEntries: DurableTaskAwaitedCompletedEntry[] = [];
  	for (const key of this._pendingCallbacks.keys()) {
  	const parts = key.split(':');
  	waitingEntries.push({
  	durableTaskExternalId: parts[0],
  	invocationCount: parseInt(parts[1], 10),
  	branchId: parseInt(parts[2], 10),
  	nodeId: parseInt(parts[3], 10),
  	});
  	}
  	this._enqueueRequest({
  	workerStatus: {
  	workerId: this._workerId,
  	waitingEntries,
  	} as DurableTaskWorkerStatusRequest,
  	});
  	}

Fixes #4445

Type of change

• Bug fix (non-breaking change which fixes an issue)
• Test changes (add, refactor, improve or change a test)

What's Changed

• Report pending callback waiting entries after reconnect and while waits remain pending.
• Add one deterministic reconnect regression test.

Checklist

Changes have been:

• Tested (unit, integration, or manually with steps specified)
• Linted and formatted
• Documented (where applicable)
• Added to CHANGELOG (where applicable) -- see Keep a Changelog

Testing

• go test -race ./pkg/client -count=1
• go test ./sdks/go/... -count=1
• gofmt -d pkg/client/durable_task_listener.go pkg/client/durable_task_listener_test.go
• git diff --check -- pkg/client/durable_task_listener.go pkg/client/durable_task_listener_test.go

---

🤖 AI Disclosure

• I acknowledge that an LLM was used in the creation of this Pull Request, in accordance with Hatchet's AI_POLICY.md.

• Details: Cursor with GPT-5.6 Sol was used to develop and test the implementation and draft this pull request.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
hatchet-docs	Ready	Preview, Comment	Jul 20, 2026 1:42pm

[github-actions]

⚠️ Optional test failure: The load-deadlock job failed on this PR (optimistic-scheduling=false). This check is non-mandatory and does not block merging, but may be worth investigating. View logs

[github-actions]

⚠️ Optional test failure: The load-deadlock job failed on this PR (optimistic-scheduling=true). This check is non-mandatory and does not block merging, but may be worth investigating. View logs

[github-actions]

⚠️ Optional test failure: The load-deadlock job failed on this PR (optimistic-scheduling=true). This check is non-mandatory and does not block merging, but may be worth investigating. View logs