This repository has been archived by the owner on Mar 5, 2024. It is now read-only.
forked from apache/datafusion-ballista
-
Notifications
You must be signed in to change notification settings - Fork 2
/
mod.rs
842 lines (736 loc) · 28.8 KB
/
mod.rs
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
// Licensed to the Apache Software Foundation (ASF) under one
// or more contributor license agreements. See the NOTICE file
// distributed with this work for additional information
// regarding copyright ownership. The ASF licenses this file
// to you under the Apache License, Version 2.0 (the
// "License"); you may not use this file except in compliance
// with the License. You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing,
// software distributed under the License is distributed on an
// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, either express or implied. See the License for the
// specific language governing permissions and limitations
// under the License.
use std::sync::Arc;
use std::time::{Duration, SystemTime, UNIX_EPOCH};
use ballista_core::config::TaskSchedulingPolicy;
use ballista_core::error::Result;
use ballista_core::event_loop::{EventLoop, EventSender};
use ballista_core::serde::protobuf::{StopExecutorParams, TaskStatus};
use ballista_core::serde::{AsExecutionPlan, BallistaCodec};
use datafusion::execution::context::{default_session_builder, SessionState};
use datafusion::logical_plan::LogicalPlan;
use datafusion::prelude::{SessionConfig, SessionContext};
use datafusion_proto::logical_plan::AsLogicalPlan;
use log::{error, warn};
use crate::scheduler_server::event::QueryStageSchedulerEvent;
use crate::scheduler_server::query_stage_scheduler::QueryStageScheduler;
use crate::state::backend::StateBackendClient;
use crate::state::executor_manager::{
ExecutorManager, ExecutorReservation, DEFAULT_EXECUTOR_TIMEOUT_SECONDS,
};
use crate::state::SchedulerState;
// include the generated protobuf source as a submodule
#[allow(clippy::all)]
pub mod externalscaler {
include!(concat!(env!("OUT_DIR"), "/externalscaler.rs"));
}
pub mod event;
mod external_scaler;
mod grpc;
mod query_stage_scheduler;
pub(crate) type SessionBuilder = fn(SessionConfig) -> SessionState;
#[derive(Clone)]
pub struct SchedulerServer<T: 'static + AsLogicalPlan, U: 'static + AsExecutionPlan> {
pub scheduler_name: String,
pub advertise_host: Option<String>,
pub(crate) state: Arc<SchedulerState<T, U>>,
pub start_time: u128,
policy: TaskSchedulingPolicy,
pub(crate) query_stage_event_loop: EventLoop<QueryStageSchedulerEvent>,
}
impl<T: 'static + AsLogicalPlan, U: 'static + AsExecutionPlan> SchedulerServer<T, U> {
pub fn new(
scheduler_name: String,
config: Arc<dyn StateBackendClient>,
codec: BallistaCodec<T, U>,
advertise_host: Option<String>,
) -> Self {
SchedulerServer::new_with_policy(
scheduler_name,
config,
TaskSchedulingPolicy::PullStaged,
codec,
default_session_builder,
advertise_host
)
}
pub fn new_with_builder(
scheduler_name: String,
config: Arc<dyn StateBackendClient>,
codec: BallistaCodec<T, U>,
session_builder: SessionBuilder,
advertise_host: Option<String>,
) -> Self {
SchedulerServer::new_with_policy(
scheduler_name,
config,
TaskSchedulingPolicy::PullStaged,
codec,
session_builder,
advertise_host
)
}
pub fn new_with_policy(
scheduler_name: String,
config: Arc<dyn StateBackendClient>,
policy: TaskSchedulingPolicy,
codec: BallistaCodec<T, U>,
session_builder: SessionBuilder,
advertise_host: Option<String>,
) -> Self {
let state = Arc::new(SchedulerState::new(
config,
session_builder,
codec,
scheduler_name.clone(),
));
SchedulerServer::new_with_state(scheduler_name, policy, state, advertise_host)
}
pub(crate) fn new_with_state(
scheduler_name: String,
policy: TaskSchedulingPolicy,
state: Arc<SchedulerState<T, U>>,
advertise_host: Option<String>,
) -> Self {
let query_stage_scheduler =
Arc::new(QueryStageScheduler::new(state.clone(), policy));
let query_stage_event_loop =
EventLoop::new("query_stage".to_owned(), 10000, query_stage_scheduler);
Self {
scheduler_name,
state,
start_time: SystemTime::now()
.duration_since(UNIX_EPOCH)
.unwrap()
.as_millis(),
policy,
query_stage_event_loop,
advertise_host,
}
}
pub async fn init(&mut self) -> Result<()> {
self.state.init().await?;
self.query_stage_event_loop.start()?;
self.expire_dead_executors()?;
Ok(())
}
pub(crate) async fn submit_job(
&self,
job_id: &str,
ctx: Arc<SessionContext>,
plan: &LogicalPlan,
) -> Result<()> {
self.query_stage_event_loop
.get_sender()?
.post_event(QueryStageSchedulerEvent::JobQueued {
job_id: job_id.to_owned(),
session_ctx: ctx,
plan: Box::new(plan.clone()),
})
.await
}
/// It just send task status update event to the channel,
/// and will not guarantee the event processing completed after return
pub(crate) async fn update_task_status(
&self,
executor_id: &str,
tasks_status: Vec<TaskStatus>,
) -> Result<()> {
// We might receive buggy task updates from dead executors.
if self.state.executor_manager.is_dead_executor(executor_id) {
let error_msg = format!(
"Receive buggy tasks status from dead Executor {}, task status update ignored.",
executor_id
);
warn!("{}", error_msg);
return Ok(());
}
self.query_stage_event_loop
.get_sender()?
.post_event(QueryStageSchedulerEvent::TaskUpdating(
executor_id.to_owned(),
tasks_status,
))
.await
}
pub(crate) async fn offer_reservation(
&self,
reservations: Vec<ExecutorReservation>,
) -> Result<()> {
self.query_stage_event_loop
.get_sender()?
.post_event(QueryStageSchedulerEvent::ReservationOffering(reservations))
.await
}
/// Spawn an async task which periodically check the active executors' status and
/// expire the dead executors
fn expire_dead_executors(&self) -> Result<()> {
let state = self.state.clone();
let event_sender = self.query_stage_event_loop.get_sender()?;
tokio::task::spawn(async move {
loop {
let expired_executors = state.executor_manager.get_expired_executors();
for expired in expired_executors {
let executor_id = expired.executor_id.clone();
let executor_manager = state.executor_manager.clone();
let stop_reason = format!(
"Executor {} heartbeat timed out after {}s",
executor_id.clone(),
DEFAULT_EXECUTOR_TIMEOUT_SECONDS
);
warn!("{}", stop_reason.clone());
let sender_clone = event_sender.clone();
Self::remove_executor(
executor_manager,
sender_clone,
&executor_id,
Some(stop_reason.clone()),
)
.await
.unwrap_or_else(|e| {
let msg = format!(
"Error to remove Executor in Scheduler due to {:?}",
e
);
error!("{}", msg);
});
match state.executor_manager.get_client(&executor_id).await {
Ok(mut client) => {
tokio::task::spawn(async move {
match client
.stop_executor(StopExecutorParams {
reason: stop_reason,
force: true,
})
.await
{
Err(error) => {
warn!(
"Failed to send stop_executor rpc due to, {}",
error
);
}
Ok(_value) => {}
}
});
}
Err(_) => {
warn!("Executor is already dead, failed to connect to Executor {}", executor_id);
}
}
}
tokio::time::sleep(Duration::from_secs(DEFAULT_EXECUTOR_TIMEOUT_SECONDS))
.await;
}
});
Ok(())
}
pub(crate) async fn remove_executor(
executor_manager: ExecutorManager,
event_sender: EventSender<QueryStageSchedulerEvent>,
executor_id: &str,
reason: Option<String>,
) -> Result<()> {
// Update the executor manager immediately here
executor_manager
.remove_executor(executor_id, reason.clone())
.await?;
event_sender
.post_event(QueryStageSchedulerEvent::ExecutorLost(
executor_id.to_owned(),
reason,
))
.await?;
Ok(())
}
}
#[cfg(all(test, feature = "sled"))]
mod test {
use std::sync::Arc;
use std::time::Duration;
use datafusion::arrow::datatypes::{DataType, Field, Schema};
use datafusion::execution::context::default_session_builder;
use datafusion::logical_plan::{col, sum, LogicalPlan};
use datafusion::test_util::scan_empty;
use datafusion_proto::protobuf::LogicalPlanNode;
use ballista_core::config::{
BallistaConfig, TaskSchedulingPolicy, BALLISTA_DEFAULT_SHUFFLE_PARTITIONS,
};
use ballista_core::error::Result;
use ballista_core::serde::protobuf::{
job_status, task_status, CompletedTask, FailedTask, JobStatus, PartitionId,
PhysicalPlanNode, ShuffleWritePartition, TaskStatus,
};
use ballista_core::serde::scheduler::{
ExecutorData, ExecutorMetadata, ExecutorSpecification,
};
use ballista_core::serde::BallistaCodec;
use crate::scheduler_server::SchedulerServer;
use crate::state::backend::standalone::StandaloneClient;
use crate::state::executor_manager::ExecutorReservation;
use crate::state::SchedulerState;
use crate::test_utils::{await_condition, ExplodingTableProvider};
#[tokio::test]
async fn test_pull_scheduling() -> Result<()> {
let plan = test_plan();
let task_slots = 4;
let scheduler = test_scheduler(TaskSchedulingPolicy::PullStaged).await?;
let executors = test_executors(task_slots);
for (executor_metadata, executor_data) in executors {
scheduler
.state
.executor_manager
.register_executor(executor_metadata, executor_data, false)
.await?;
}
let config = test_session(task_slots);
let ctx = scheduler
.state
.session_manager
.create_session(&config)
.await?;
let job_id = "job";
// Submit job
scheduler
.state
.submit_job(job_id, ctx, &plan)
.await
.expect("submitting plan");
// Refresh the ExecutionGraph
while let Some(graph) = scheduler
.state
.task_manager
.get_active_execution_graph(job_id)
.await
{
let task = {
let mut graph = graph.write().await;
graph.pop_next_task("executor-1")?
};
if let Some(task) = task {
let mut partitions: Vec<ShuffleWritePartition> = vec![];
let num_partitions = task
.output_partitioning
.map(|p| p.partition_count())
.unwrap_or(1);
for partition_id in 0..num_partitions {
partitions.push(ShuffleWritePartition {
partition_id: partition_id as u64,
path: "some/path".to_string(),
num_batches: 1,
num_rows: 1,
num_bytes: 1,
})
}
// Complete the task
let task_status = TaskStatus {
status: Some(task_status::Status::Completed(CompletedTask {
executor_id: "executor-1".to_owned(),
partitions,
})),
metrics: vec![],
task_id: Some(PartitionId {
job_id: job_id.to_owned(),
stage_id: task.partition.stage_id as u32,
partition_id: task.partition.partition_id as u32,
}),
};
scheduler
.state
.update_task_statuses("executor-1", vec![task_status])
.await?;
} else {
break;
}
}
let final_graph = scheduler
.state
.task_manager
.get_active_execution_graph(job_id)
.await
.expect("Fail to find graph in the cache");
let final_graph = final_graph.read().await;
assert!(final_graph.complete());
assert_eq!(final_graph.output_locations().len(), 4);
for output_location in final_graph.output_locations() {
assert_eq!(output_location.path, "some/path".to_owned());
assert_eq!(output_location.executor_meta.host, "localhost1".to_owned())
}
Ok(())
}
/// This test will exercise the push-based scheduling.
#[tokio::test]
async fn test_push_scheduling() -> Result<()> {
let plan = test_plan();
let task_slots = 4;
let scheduler = test_push_staged_scheduler().await?;
let executors = test_executors(task_slots);
for (executor_metadata, executor_data) in executors {
scheduler
.state
.executor_manager
.register_executor(executor_metadata, executor_data, false)
.await?;
}
let config = test_session(task_slots);
let ctx = scheduler
.state
.session_manager
.create_session(&config)
.await?;
let job_id = "job";
scheduler.state.submit_job(job_id, ctx, &plan).await?;
// Complete tasks that are offered through scheduler events
loop {
// Check condition
let available_tasks = {
let graph = scheduler
.state
.task_manager
.get_active_execution_graph(job_id)
.await
.unwrap();
let graph = graph.read().await;
if graph.complete() {
break;
}
graph.available_tasks()
};
if available_tasks == 0 {
tokio::time::sleep(Duration::from_millis(5)).await;
continue;
}
let reservations: Vec<ExecutorReservation> = scheduler
.state
.executor_manager
.reserve_slots(available_tasks as u32)
.await?
.into_iter()
.map(|res| res.assign(job_id.to_owned()))
.collect();
let free_list = match scheduler
.state
.task_manager
.fill_reservations(&reservations)
.await
{
Ok((assignments, mut unassigned_reservations, _)) => {
for (executor_id, task) in assignments.into_iter() {
match scheduler
.state
.executor_manager
.get_executor_metadata(&executor_id)
.await
{
Ok(executor) => {
let mut partitions: Vec<ShuffleWritePartition> = vec![];
let num_partitions = task
.output_partitioning
.map(|p| p.partition_count())
.unwrap_or(1);
for partition_id in 0..num_partitions {
partitions.push(ShuffleWritePartition {
partition_id: partition_id as u64,
path: "some/path".to_string(),
num_batches: 1,
num_rows: 1,
num_bytes: 1,
})
}
// Complete the task
let task_status = TaskStatus {
status: Some(task_status::Status::Completed(
CompletedTask {
executor_id: executor.id.clone(),
partitions,
},
)),
metrics: vec![],
task_id: Some(PartitionId {
job_id: job_id.to_owned(),
stage_id: task.partition.stage_id as u32,
partition_id: task.partition.partition_id as u32,
}),
};
scheduler
.update_task_status(&executor.id, vec![task_status])
.await?;
}
Err(_e) => {
unassigned_reservations.push(
ExecutorReservation::new_free(executor_id.clone()),
);
}
}
}
unassigned_reservations
}
Err(_e) => reservations,
};
// If any reserved slots remain, return them to the pool
if !free_list.is_empty() {
scheduler
.state
.executor_manager
.cancel_reservations(free_list)
.await?;
}
}
let final_graph = scheduler
.state
.task_manager
.get_execution_graph(job_id)
.await?;
assert!(final_graph.complete());
assert_eq!(final_graph.output_locations().len(), 4);
Ok(())
}
// Simulate a task failure and ensure the job status is updated correctly
#[tokio::test]
async fn test_job_failure() -> Result<()> {
let plan = test_plan();
let task_slots = 4;
let scheduler = test_push_staged_scheduler().await?;
let executors = test_executors(task_slots);
for (executor_metadata, executor_data) in executors {
scheduler
.state
.executor_manager
.register_executor(executor_metadata, executor_data, false)
.await?;
}
let config = test_session(task_slots);
let ctx = scheduler
.state
.session_manager
.create_session(&config)
.await?;
let job_id = "job";
scheduler.state.submit_job(job_id, ctx, &plan).await?;
let available_tasks = scheduler
.state
.task_manager
.get_available_task_count(job_id)
.await?;
let reservations: Vec<ExecutorReservation> = scheduler
.state
.executor_manager
.reserve_slots(available_tasks as u32)
.await?
.into_iter()
.map(|res| res.assign(job_id.to_owned()))
.collect();
// Complete tasks that are offered through scheduler events
let free_list = match scheduler
.state
.task_manager
.fill_reservations(&reservations)
.await
{
Ok((assignments, mut unassigned_reservations, _)) => {
for (executor_id, task) in assignments.into_iter() {
match scheduler
.state
.executor_manager
.get_executor_metadata(&executor_id)
.await
{
Ok(executor) => {
let mut partitions: Vec<ShuffleWritePartition> = vec![];
let num_partitions = task
.output_partitioning
.map(|p| p.partition_count())
.unwrap_or(1);
for partition_id in 0..num_partitions {
partitions.push(ShuffleWritePartition {
partition_id: partition_id as u64,
path: "some/path".to_string(),
num_batches: 1,
num_rows: 1,
num_bytes: 1,
})
}
// Complete the task
let task_status = TaskStatus {
status: Some(task_status::Status::Failed(FailedTask {
error: "".to_string(),
})),
metrics: vec![],
task_id: Some(PartitionId {
job_id: job_id.to_owned(),
stage_id: task.partition.stage_id as u32,
partition_id: task.partition.partition_id as u32,
}),
};
scheduler
.state
.update_task_statuses(&executor.id, vec![task_status])
.await?;
}
Err(_e) => {
unassigned_reservations
.push(ExecutorReservation::new_free(executor_id.clone()));
}
}
}
unassigned_reservations
}
Err(_e) => reservations,
};
// If any reserved slots remain, return them to the pool
if !free_list.is_empty() {
scheduler
.state
.executor_manager
.cancel_reservations(free_list)
.await?;
}
let status = scheduler.state.task_manager.get_job_status(job_id).await?;
assert!(
matches!(
status,
Some(JobStatus {
status: Some(job_status::Status::Failed(_))
})
),
"Expected job status to be failed"
);
Ok(())
}
// If the physical planning fails, the job should be marked as failed.
// Here we simulate a planning failure using ExplodingTableProvider to test this.
#[tokio::test]
async fn test_planning_failure() -> Result<()> {
let task_slots = 4;
let scheduler = test_push_staged_scheduler().await?;
let config = test_session(task_slots);
let ctx = scheduler
.state
.session_manager
.create_session(&config)
.await?;
ctx.register_table("explode", Arc::new(ExplodingTableProvider))?;
let plan = ctx.sql("SELECT * FROM explode").await?.to_logical_plan()?;
let job_id = "job";
// This should fail when we try and create the physical plan
scheduler.submit_job(job_id, ctx, &plan).await?;
let scheduler = scheduler.clone();
let check = || async {
let status = scheduler.state.task_manager.get_job_status(job_id).await?;
Ok(matches!(
status,
Some(JobStatus {
status: Some(job_status::Status::Failed(_))
})
))
};
// Sine this happens in an event loop, we need to check a few times.
let job_failed = await_condition(Duration::from_millis(100), 10, check).await?;
assert!(job_failed, "Job status not failed after 1 second");
Ok(())
}
async fn test_scheduler(
policy: TaskSchedulingPolicy,
) -> Result<SchedulerServer<LogicalPlanNode, PhysicalPlanNode>> {
let state_storage = Arc::new(StandaloneClient::try_new_temporary()?);
let mut scheduler: SchedulerServer<LogicalPlanNode, PhysicalPlanNode> =
SchedulerServer::new_with_policy(
"localhost:50050".to_owned(),
state_storage.clone(),
policy,
BallistaCodec::default(),
default_session_builder,
None,
);
scheduler.init().await?;
Ok(scheduler)
}
async fn test_push_staged_scheduler(
) -> Result<SchedulerServer<LogicalPlanNode, PhysicalPlanNode>> {
let state_storage = Arc::new(StandaloneClient::try_new_temporary()?);
let state = Arc::new(SchedulerState::new_with_default_scheduler_name(
state_storage,
default_session_builder,
BallistaCodec::default(),
));
let mut scheduler: SchedulerServer<LogicalPlanNode, PhysicalPlanNode> =
SchedulerServer::new_with_state(
"localhost:50050".to_owned(),
TaskSchedulingPolicy::PushStaged,
state,
None
);
scheduler.init().await?;
Ok(scheduler)
}
fn test_executors(num_partitions: usize) -> Vec<(ExecutorMetadata, ExecutorData)> {
let task_slots = (num_partitions as u32 + 1) / 2;
vec![
(
ExecutorMetadata {
id: "executor-1".to_string(),
host: "localhost1".to_string(),
port: 8080,
grpc_port: 9090,
specification: ExecutorSpecification { task_slots },
},
ExecutorData {
executor_id: "executor-1".to_owned(),
total_task_slots: task_slots,
available_task_slots: task_slots,
},
),
(
ExecutorMetadata {
id: "executor-2".to_string(),
host: "localhost2".to_string(),
port: 8080,
grpc_port: 9090,
specification: ExecutorSpecification {
task_slots: num_partitions as u32 - task_slots,
},
},
ExecutorData {
executor_id: "executor-2".to_owned(),
total_task_slots: num_partitions as u32 - task_slots,
available_task_slots: num_partitions as u32 - task_slots,
},
),
]
}
fn test_plan() -> LogicalPlan {
let schema = Schema::new(vec![
Field::new("id", DataType::Utf8, false),
Field::new("gmv", DataType::UInt64, false),
]);
scan_empty(None, &schema, Some(vec![0, 1]))
.unwrap()
.aggregate(vec![col("id")], vec![sum(col("gmv"))])
.unwrap()
.build()
.unwrap()
}
fn test_session(partitions: usize) -> BallistaConfig {
BallistaConfig::builder()
.set(
BALLISTA_DEFAULT_SHUFFLE_PARTITIONS,
format!("{}", partitions).as_str(),
)
.build()
.expect("creating BallistaConfig")
}
}