async fn render_no_shuffle_ensembles<C>(
txn: &C,
actor_id_counter: &AtomicU32,
ensembles: &[NoShuffleEnsemble],
fragment_map: &HashMap<FragmentId, Model>,
job_map: &HashMap<JobId, Model>,
worker_map: &BTreeMap<WorkerId, WorkerInfo>,
adaptive_parallelism_strategy: AdaptiveParallelismStrategy,
) -> MetaResult<FragmentRenderMap>where
C: ConnectionTrait,Expand description
Core rendering routine that consumes no-shuffle ensembles and produces
InflightFragmentInfos by:
- determining the eligible worker pools and effective parallelism,
- generating actor→worker assignments plus vnode bitmaps and source splits,
- grouping the rendered fragments by database and streaming job.