I’m doing manual cursor management in a sensor. The sensor is appropriately triggering the target pipeline in my dev environment but not my stage environment. I don’t understand how that can be possible. Here’s the relevant sensor code:

@sensor(pipeline_name="collective_pipeline", mode="stage_coll")
def collective_sensor(context):
    """ Workaround for wildcard matching on asset_keys. From <https://github.com/dagster-io/dagster/issues/4582> """
    after_cursor = int(context.cursor) if context.cursor else 0
    event_records = context.instance.get_event_records(
        EventRecordsFilter(
            event_type=DagsterEventType.ASSET_MATERIALIZATION,
            after_cursor=after_cursor,
        ),
        ascending=True,
    )

    for event_record in event_records:
        ...

    if len(event_records) > 0:
        context.update_cursor(str(event_records[0].storage_id))
    else:
        yield SkipReason(f"No Asset materialization event records found. Cursor: {after_cursor}")

The sensor will, every 30 seconds, reach the last line and produce a skip result saying

No Asset materialization event records found. Cursor: 0

I can see all the asset materializations in the Asset tab just fine. my dev environment uses the

DockerRunLauncher

and

LocalComputeLogManager

while my stage uses the

EcsRunLauncher

and

S3ComputeLogManager

respectively. Other than that, the two environments should be pretty much the same.