Not exactly sure how this code path gets exercised and why test case like https://github.com/flyteorg/flytekit/blob/b6605bc95bd6f03434aeb5ca8f69eba161924ef8/tests/flytekit/integration/remote/test_remote.py#L219 does not catch it.

I guess the fix would be like the following?

task_node_exec[0].id.node_id: FlyteNode(
                            id=flyte_entity.id,
                            upstream_nodes=[],
                            bindings=[],
                            metadata=NodeMetadata(name=""),
                            task_node=FlyteTaskNode(flyte_entity),
                        )

that is correct

A question. Why

wait=True

does not result

execution = self.sync_execution(execution, sync_nodes=sync_nodes)

being called in

wait

function.

Our test case does pretty much the same thing. Not sure why it breaks for us. 🤔 This is the stack trace for us:

tests/test_util/remote_util.py:6: in remote_handle
    execution = remote.execute(
/Users/honnix/.pyenv/versions/spotify-flytekit/lib/python3.8/site-packages/flytekit/remote/remote.py:1051: in execute
    return self.execute_remote_task_lp(
/Users/honnix/.pyenv/versions/spotify-flytekit/lib/python3.8/site-packages/flytekit/remote/remote.py:1134: in execute_remote_task_lp
    return self._execute(
/Users/honnix/.pyenv/versions/spotify-flytekit/lib/python3.8/site-packages/flytekit/remote/remote.py:964: in _execute
    return self.wait(execution)
/Users/honnix/.pyenv/versions/spotify-flytekit/lib/python3.8/site-packages/flytekit/remote/remote.py:1381: in wait
    execution = self.sync_execution(execution, sync_nodes=sync_nodes)

So setting

wait=True

is sufficient to catch this in our case.

Do we have a plan to drop a 1.3.x release to include this fix? Thanks.