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
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166 | @register_base_type
class BaseWorker(abc.ABC):
type: str
job_configuration: Type[BaseJobConfiguration] = BaseJobConfiguration
job_configuration_variables: Optional[Type[BaseVariables]] = None
_documentation_url = ""
_logo_url = ""
_description = ""
@experimental(feature="The workers feature", group="workers")
def __init__(
self,
work_pool_name: str,
work_queues: Optional[List[str]] = None,
name: Optional[str] = None,
prefetch_seconds: Optional[float] = None,
create_pool_if_not_found: bool = True,
limit: Optional[int] = None,
heartbeat_interval_seconds: Optional[int] = None,
*,
base_job_template: Optional[Dict[str, Any]] = None,
):
"""
Base class for all Prefect workers.
Args:
name: The name of the worker. If not provided, a random one
will be generated. If provided, it cannot contain '/' or '%'.
The name is used to identify the worker in the UI; if two
processes have the same name, they will be treated as the same
worker.
work_pool_name: The name of the work pool to poll.
work_queues: A list of work queues to poll. If not provided, all
work queue in the work pool will be polled.
prefetch_seconds: The number of seconds to prefetch flow runs for.
create_pool_if_not_found: Whether to create the work pool
if it is not found. Defaults to `True`, but can be set to `False` to
ensure that work pools are not created accidentally.
limit: The maximum number of flow runs this worker should be running at
a given time.
base_job_template: If creating the work pool, provide the base job
template to use. Logs a warning if the pool already exists.
"""
if name and ("/" in name or "%" in name):
raise ValueError("Worker name cannot contain '/' or '%'")
self.name = name or f"{self.__class__.__name__} {uuid4()}"
self._logger = get_logger(f"worker.{self.__class__.type}.{self.name.lower()}")
self.is_setup = False
self._create_pool_if_not_found = create_pool_if_not_found
self._base_job_template = base_job_template
self._work_pool_name = work_pool_name
self._work_queues: Set[str] = set(work_queues) if work_queues else set()
self._prefetch_seconds: float = (
prefetch_seconds or PREFECT_WORKER_PREFETCH_SECONDS.value()
)
self.heartbeat_interval_seconds = (
heartbeat_interval_seconds or PREFECT_WORKER_HEARTBEAT_SECONDS.value()
)
self._work_pool: Optional[WorkPool] = None
self._runs_task_group: Optional[anyio.abc.TaskGroup] = None
self._client: Optional[PrefectClient] = None
self._last_polled_time: pendulum.DateTime = pendulum.now("utc")
self._limit = limit
self._limiter: Optional[anyio.CapacityLimiter] = None
self._submitting_flow_run_ids = set()
self._cancelling_flow_run_ids = set()
self._scheduled_task_scopes = set()
@classmethod
def get_documentation_url(cls) -> str:
return cls._documentation_url
@classmethod
def get_logo_url(cls) -> str:
return cls._logo_url
@classmethod
def get_description(cls) -> str:
return cls._description
@classmethod
def get_default_base_job_template(cls) -> Dict:
if cls.job_configuration_variables is None:
schema = cls.job_configuration.schema()
# remove "template" key from all dicts in schema['properties'] because it is not a
# relevant field
for key, value in schema["properties"].items():
if isinstance(value, dict):
schema["properties"][key].pop("template", None)
variables_schema = schema
else:
variables_schema = cls.job_configuration_variables.schema()
variables_schema.pop("title", None)
return {
"job_configuration": cls.job_configuration.json_template(),
"variables": variables_schema,
}
@staticmethod
def get_worker_class_from_type(type: str) -> Optional[Type["BaseWorker"]]:
"""
Returns the worker class for a given worker type. If the worker type
is not recognized, returns None.
"""
load_prefect_collections()
worker_registry = get_registry_for_type(BaseWorker)
if worker_registry is not None:
return worker_registry.get(type)
@staticmethod
def get_all_available_worker_types() -> List[str]:
"""
Returns all worker types available in the local registry.
"""
load_prefect_collections()
worker_registry = get_registry_for_type(BaseWorker)
if worker_registry is not None:
return list(worker_registry.keys())
return []
def get_name_slug(self):
return slugify(self.name)
def get_flow_run_logger(self, flow_run: "FlowRun") -> PrefectLogAdapter:
return flow_run_logger(flow_run=flow_run).getChild(
"worker",
extra={
"worker_name": self.name,
"work_pool_name": (
self._work_pool_name if self._work_pool else "<unknown>"
),
"work_pool_id": str(getattr(self._work_pool, "id", "unknown")),
},
)
@abc.abstractmethod
async def run(
self,
flow_run: "FlowRun",
configuration: BaseJobConfiguration,
task_status: Optional[anyio.abc.TaskStatus] = None,
) -> BaseWorkerResult:
"""
Runs a given flow run on the current worker.
"""
raise NotImplementedError(
"Workers must implement a method for running submitted flow runs"
)
async def kill_infrastructure(
self,
infrastructure_pid: str,
configuration: BaseJobConfiguration,
grace_seconds: int = 30,
):
"""
Method for killing infrastructure created by a worker. Should be implemented by
individual workers if they support killing infrastructure.
"""
raise NotImplementedError(
"This worker does not support killing infrastructure."
)
@classmethod
def __dispatch_key__(cls):
if cls.__name__ == "BaseWorker":
return None # The base class is abstract
return cls.type
async def setup(self):
"""Prepares the worker to run."""
self._logger.debug("Setting up worker...")
self._runs_task_group = anyio.create_task_group()
self._limiter = (
anyio.CapacityLimiter(self._limit) if self._limit is not None else None
)
self._client = get_client()
await self._client.__aenter__()
await self._runs_task_group.__aenter__()
self.is_setup = True
async def teardown(self, *exc_info):
"""Cleans up resources after the worker is stopped."""
self._logger.debug("Tearing down worker...")
self.is_setup = False
for scope in self._scheduled_task_scopes:
scope.cancel()
if self._runs_task_group:
await self._runs_task_group.__aexit__(*exc_info)
if self._client:
await self._client.__aexit__(*exc_info)
self._runs_task_group = None
self._client = None
def is_worker_still_polling(self, query_interval_seconds: int) -> bool:
"""
This method is invoked by a webserver healthcheck handler
and returns a boolean indicating if the worker has recorded a
scheduled flow run poll within a variable amount of time.
The `query_interval_seconds` is the same value that is used by
the loop services - we will evaluate if the _last_polled_time
was within that interval x 30 (so 10s -> 5m)
The instance property `self._last_polled_time`
is currently set/updated in `get_and_submit_flow_runs()`
"""
threshold_seconds = query_interval_seconds * 30
seconds_since_last_poll = (
pendulum.now("utc") - self._last_polled_time
).in_seconds()
is_still_polling = seconds_since_last_poll <= threshold_seconds
if not is_still_polling:
self._logger.error(
f"Worker has not polled in the last {seconds_since_last_poll} seconds "
"and should be restarted"
)
return is_still_polling
async def get_and_submit_flow_runs(self):
runs_response = await self._get_scheduled_flow_runs()
self._last_polled_time = pendulum.now("utc")
return await self._submit_scheduled_flow_runs(flow_run_response=runs_response)
async def check_for_cancelled_flow_runs(self):
if not self.is_setup:
raise RuntimeError(
"Worker is not set up. Please make sure you are running this worker "
"as an async context manager."
)
self._logger.debug("Checking for cancelled flow runs...")
work_queue_filter = (
WorkQueueFilter(name=WorkQueueFilterName(any_=list(self._work_queues)))
if self._work_queues
else None
)
named_cancelling_flow_runs = await self._client.read_flow_runs(
flow_run_filter=FlowRunFilter(
state=FlowRunFilterState(
type=FlowRunFilterStateType(any_=[StateType.CANCELLED]),
name=FlowRunFilterStateName(any_=["Cancelling"]),
),
# Avoid duplicate cancellation calls
id=FlowRunFilterId(not_any_=list(self._cancelling_flow_run_ids)),
),
work_pool_filter=WorkPoolFilter(
name=WorkPoolFilterName(any_=[self._work_pool_name])
),
work_queue_filter=work_queue_filter,
)
typed_cancelling_flow_runs = await self._client.read_flow_runs(
flow_run_filter=FlowRunFilter(
state=FlowRunFilterState(
type=FlowRunFilterStateType(any_=[StateType.CANCELLING]),
),
# Avoid duplicate cancellation calls
id=FlowRunFilterId(not_any_=list(self._cancelling_flow_run_ids)),
),
work_pool_filter=WorkPoolFilter(
name=WorkPoolFilterName(any_=[self._work_pool_name])
),
work_queue_filter=work_queue_filter,
)
cancelling_flow_runs = named_cancelling_flow_runs + typed_cancelling_flow_runs
if cancelling_flow_runs:
self._logger.info(
f"Found {len(cancelling_flow_runs)} flow runs awaiting cancellation."
)
for flow_run in cancelling_flow_runs:
self._cancelling_flow_run_ids.add(flow_run.id)
self._runs_task_group.start_soon(self.cancel_run, flow_run)
return cancelling_flow_runs
async def cancel_run(self, flow_run: "FlowRun"):
run_logger = self.get_flow_run_logger(flow_run)
if not flow_run.infrastructure_pid:
run_logger.error(
f"Flow run '{flow_run.id}' does not have an infrastructure pid"
" attached. Cancellation cannot be guaranteed."
)
await self._mark_flow_run_as_cancelled(
flow_run,
state_updates={
"message": (
"This flow run is missing infrastructure tracking information"
" and cancellation cannot be guaranteed."
)
},
)
return
try:
configuration = await self._get_configuration(flow_run)
except ObjectNotFound:
self._logger.warning(
f"Flow run {flow_run.id!r} cannot be cancelled by this worker:"
f" associated deployment {flow_run.deployment_id!r} does not exist."
)
try:
await self.kill_infrastructure(
infrastructure_pid=flow_run.infrastructure_pid,
configuration=configuration,
)
except NotImplementedError:
self._logger.error(
f"Worker type {self.type!r} does not support killing created "
"infrastructure. Cancellation cannot be guaranteed."
)
except InfrastructureNotFound as exc:
self._logger.warning(f"{exc} Marking flow run as cancelled.")
await self._mark_flow_run_as_cancelled(flow_run)
except InfrastructureNotAvailable as exc:
self._logger.warning(f"{exc} Flow run cannot be cancelled by this worker.")
except Exception:
run_logger.exception(
"Encountered exception while killing infrastructure for flow run "
f"'{flow_run.id}'. Flow run may not be cancelled."
)
# We will try again on generic exceptions
self._cancelling_flow_run_ids.remove(flow_run.id)
return
else:
self._emit_flow_run_cancelled_event(
flow_run=flow_run, configuration=configuration
)
await self._mark_flow_run_as_cancelled(flow_run)
run_logger.info(f"Cancelled flow run '{flow_run.id}'!")
async def _update_local_work_pool_info(self):
try:
work_pool = await self._client.read_work_pool(
work_pool_name=self._work_pool_name
)
except ObjectNotFound:
if self._create_pool_if_not_found:
wp = WorkPoolCreate(
name=self._work_pool_name,
type=self.type,
)
if self._base_job_template is not None:
wp.base_job_template = self._base_job_template
work_pool = await self._client.create_work_pool(work_pool=wp)
self._logger.info(f"Work pool {self._work_pool_name!r} created.")
else:
self._logger.warning(f"Work pool {self._work_pool_name!r} not found!")
if self._base_job_template is not None:
self._logger.warning(
"Ignoring supplied base job template because the work pool"
" already exists"
)
return
# if the remote config type changes (or if it's being loaded for the
# first time), check if it matches the local type and warn if not
if getattr(self._work_pool, "type", 0) != work_pool.type:
if work_pool.type != self.__class__.type:
self._logger.warning(
"Worker type mismatch! This worker process expects type "
f"{self.type!r} but received {work_pool.type!r}"
" from the server. Unexpected behavior may occur."
)
# once the work pool is loaded, verify that it has a `base_job_template` and
# set it if not
if not work_pool.base_job_template:
job_template = self.__class__.get_default_base_job_template()
await self._set_work_pool_template(work_pool, job_template)
work_pool.base_job_template = job_template
self._work_pool = work_pool
async def _send_worker_heartbeat(self):
if self._work_pool:
await self._client.send_worker_heartbeat(
work_pool_name=self._work_pool_name,
worker_name=self.name,
heartbeat_interval_seconds=self.heartbeat_interval_seconds,
)
async def sync_with_backend(self):
"""
Updates the worker's local information about it's current work pool and
queues. Sends a worker heartbeat to the API.
"""
await self._update_local_work_pool_info()
await self._send_worker_heartbeat()
self._logger.debug("Worker synchronized with the Prefect API server.")
async def _get_scheduled_flow_runs(
self,
) -> List["WorkerFlowRunResponse"]:
"""
Retrieve scheduled flow runs from the work pool's queues.
"""
scheduled_before = pendulum.now("utc").add(seconds=int(self._prefetch_seconds))
self._logger.debug(
f"Querying for flow runs scheduled before {scheduled_before}"
)
try:
scheduled_flow_runs = (
await self._client.get_scheduled_flow_runs_for_work_pool(
work_pool_name=self._work_pool_name,
scheduled_before=scheduled_before,
work_queue_names=list(self._work_queues),
)
)
self._logger.debug(
f"Discovered {len(scheduled_flow_runs)} scheduled_flow_runs"
)
return scheduled_flow_runs
except ObjectNotFound:
# the pool doesn't exist; it will be created on the next
# heartbeat (or an appropriate warning will be logged)
return []
async def _submit_scheduled_flow_runs(
self, flow_run_response: List["WorkerFlowRunResponse"]
) -> List["FlowRun"]:
"""
Takes a list of WorkerFlowRunResponses and submits the referenced flow runs
for execution by the worker.
"""
submittable_flow_runs = [entry.flow_run for entry in flow_run_response]
submittable_flow_runs.sort(key=lambda run: run.next_scheduled_start_time)
for flow_run in submittable_flow_runs:
if flow_run.id in self._submitting_flow_run_ids:
continue
try:
if self._limiter:
self._limiter.acquire_on_behalf_of_nowait(flow_run.id)
except anyio.WouldBlock:
self._logger.info(
f"Flow run limit reached; {self._limiter.borrowed_tokens} flow runs"
" in progress."
)
break
else:
run_logger = self.get_flow_run_logger(flow_run)
run_logger.info(
f"Worker '{self.name}' submitting flow run '{flow_run.id}'"
)
self._submitting_flow_run_ids.add(flow_run.id)
self._runs_task_group.start_soon(
self._submit_run,
flow_run,
)
return list(
filter(
lambda run: run.id in self._submitting_flow_run_ids,
submittable_flow_runs,
)
)
async def _check_flow_run(self, flow_run: "FlowRun") -> None:
"""
Performs a check on a submitted flow run to warn the user if the flow run
was created from a deployment with a storage block.
"""
if flow_run.deployment_id:
deployment = await self._client.read_deployment(flow_run.deployment_id)
if deployment.storage_document_id:
raise ValueError(
f"Flow run {flow_run.id!r} was created from deployment"
f" {deployment.name!r} which is configured with a storage block."
" Workers currently only support local storage. Please use an"
" agent to execute this flow run."
)
async def _submit_run(self, flow_run: "FlowRun") -> None:
"""
Submits a given flow run for execution by the worker.
"""
run_logger = self.get_flow_run_logger(flow_run)
try:
await self._check_flow_run(flow_run)
except (ValueError, ObjectNotFound):
self._logger.exception(
(
"Flow run %s did not pass checks and will not be submitted for"
" execution"
),
flow_run.id,
)
self._submitting_flow_run_ids.remove(flow_run.id)
return
ready_to_submit = await self._propose_pending_state(flow_run)
if ready_to_submit:
readiness_result = await self._runs_task_group.start(
self._submit_run_and_capture_errors, flow_run
)
if readiness_result and not isinstance(readiness_result, Exception):
try:
await self._client.update_flow_run(
flow_run_id=flow_run.id,
infrastructure_pid=str(readiness_result),
)
except Exception:
run_logger.exception(
"An error occurred while setting the `infrastructure_pid` on "
f"flow run {flow_run.id!r}. The flow run will "
"not be cancellable."
)
run_logger.info(f"Completed submission of flow run '{flow_run.id}'")
else:
# If the run is not ready to submit, release the concurrency slot
if self._limiter:
self._limiter.release_on_behalf_of(flow_run.id)
self._submitting_flow_run_ids.remove(flow_run.id)
async def _submit_run_and_capture_errors(
self, flow_run: "FlowRun", task_status: anyio.abc.TaskStatus = None
) -> Union[BaseWorkerResult, Exception]:
run_logger = self.get_flow_run_logger(flow_run)
try:
configuration = await self._get_configuration(flow_run)
submitted_event = self._emit_flow_run_submitted_event(configuration)
result = await self.run(
flow_run=flow_run,
task_status=task_status,
configuration=configuration,
)
except Exception as exc:
if not task_status._future.done():
# This flow run was being submitted and did not start successfully
run_logger.exception(
f"Failed to submit flow run '{flow_run.id}' to infrastructure."
)
# Mark the task as started to prevent agent crash
task_status.started(exc)
await self._propose_crashed_state(
flow_run, "Flow run could not be submitted to infrastructure"
)
else:
run_logger.exception(
f"An error occurred while monitoring flow run '{flow_run.id}'. "
"The flow run will not be marked as failed, but an issue may have "
"occurred."
)
return exc
finally:
if self._limiter:
self._limiter.release_on_behalf_of(flow_run.id)
if not task_status._future.done():
run_logger.error(
f"Infrastructure returned without reporting flow run '{flow_run.id}' "
"as started or raising an error. This behavior is not expected and "
"generally indicates improper implementation of infrastructure. The "
"flow run will not be marked as failed, but an issue may have occurred."
)
# Mark the task as started to prevent agent crash
task_status.started()
if result.status_code != 0:
await self._propose_crashed_state(
flow_run,
(
"Flow run infrastructure exited with non-zero status code"
f" {result.status_code}."
),
)
self._emit_flow_run_executed_event(result, configuration, submitted_event)
return result
def get_status(self):
"""
Retrieves the status of the current worker including its name, current worker
pool, the work pool queues it is polling, and its local settings.
"""
return {
"name": self.name,
"work_pool": (
self._work_pool.dict(json_compatible=True)
if self._work_pool is not None
else None
),
"settings": {
"prefetch_seconds": self._prefetch_seconds,
},
}
async def _get_configuration(
self,
flow_run: "FlowRun",
) -> BaseJobConfiguration:
deployment = await self._client.read_deployment(flow_run.deployment_id)
flow = await self._client.read_flow(flow_run.flow_id)
configuration = await self.job_configuration.from_template_and_values(
base_job_template=self._work_pool.base_job_template,
values=deployment.infra_overrides or {},
client=self._client,
)
configuration.prepare_for_flow_run(
flow_run=flow_run, deployment=deployment, flow=flow
)
return configuration
async def _propose_pending_state(self, flow_run: "FlowRun") -> bool:
run_logger = self.get_flow_run_logger(flow_run)
state = flow_run.state
try:
state = await propose_state(
self._client, Pending(), flow_run_id=flow_run.id
)
except Abort as exc:
run_logger.info(
(
f"Aborted submission of flow run '{flow_run.id}'. "
f"Server sent an abort signal: {exc}"
),
)
return False
except Exception:
run_logger.exception(
f"Failed to update state of flow run '{flow_run.id}'",
)
return False
if not state.is_pending():
run_logger.info(
(
f"Aborted submission of flow run '{flow_run.id}': "
f"Server returned a non-pending state {state.type.value!r}"
),
)
return False
return True
async def _propose_failed_state(self, flow_run: "FlowRun", exc: Exception) -> None:
run_logger = self.get_flow_run_logger(flow_run)
try:
await propose_state(
self._client,
await exception_to_failed_state(message="Submission failed.", exc=exc),
flow_run_id=flow_run.id,
)
except Abort:
# We've already failed, no need to note the abort but we don't want it to
# raise in the agent process
pass
except Exception:
run_logger.error(
f"Failed to update state of flow run '{flow_run.id}'",
exc_info=True,
)
async def _propose_crashed_state(self, flow_run: "FlowRun", message: str) -> None:
run_logger = self.get_flow_run_logger(flow_run)
try:
state = await propose_state(
self._client,
Crashed(message=message),
flow_run_id=flow_run.id,
)
except Abort:
# Flow run already marked as failed
pass
except Exception:
run_logger.exception(f"Failed to update state of flow run '{flow_run.id}'")
else:
if state.is_crashed():
run_logger.info(
f"Reported flow run '{flow_run.id}' as crashed: {message}"
)
async def _mark_flow_run_as_cancelled(
self, flow_run: "FlowRun", state_updates: Optional[dict] = None
) -> None:
state_updates = state_updates or {}
state_updates.setdefault("name", "Cancelled")
state_updates.setdefault("type", StateType.CANCELLED)
state = flow_run.state.copy(update=state_updates)
await self._client.set_flow_run_state(flow_run.id, state, force=True)
# Do not remove the flow run from the cancelling set immediately because
# the API caches responses for the `read_flow_runs` and we do not want to
# duplicate cancellations.
await self._schedule_task(
60 * 10, self._cancelling_flow_run_ids.remove, flow_run.id
)
async def _set_work_pool_template(self, work_pool, job_template):
"""Updates the `base_job_template` for the worker's work pool server side."""
await self._client.update_work_pool(
work_pool_name=work_pool.name,
work_pool=WorkPoolUpdate(
base_job_template=job_template,
),
)
async def _schedule_task(self, __in_seconds: int, fn, *args, **kwargs):
"""
Schedule a background task to start after some time.
These tasks will be run immediately when the worker exits instead of waiting.
The function may be async or sync. Async functions will be awaited.
"""
async def wrapper(task_status):
# If we are shutting down, do not sleep; otherwise sleep until the scheduled
# time or shutdown
if self.is_setup:
with anyio.CancelScope() as scope:
self._scheduled_task_scopes.add(scope)
task_status.started()
await anyio.sleep(__in_seconds)
self._scheduled_task_scopes.remove(scope)
else:
task_status.started()
result = fn(*args, **kwargs)
if inspect.iscoroutine(result):
await result
await self._runs_task_group.start(wrapper)
async def __aenter__(self):
self._logger.debug("Entering worker context...")
await self.setup()
return self
async def __aexit__(self, *exc_info):
self._logger.debug("Exiting worker context...")
await self.teardown(*exc_info)
def __repr__(self):
return f"Worker(pool={self._work_pool_name!r}, name={self.name!r})"
def _event_resource(self):
return {
"prefect.resource.id": f"prefect.worker.{self.type}.{self.get_name_slug()}",
"prefect.resource.name": self.name,
"prefect.version": prefect.__version__,
"prefect.worker-type": self.type,
}
def _event_related_resources(
self,
configuration: Optional[BaseJobConfiguration] = None,
include_self: bool = False,
) -> List[RelatedResource]:
related = []
if configuration:
related += configuration._related_resources()
if self._work_pool:
related.append(
object_as_related_resource(
kind="work-pool", role="work-pool", object=self._work_pool
)
)
if include_self:
worker_resource = self._event_resource()
worker_resource["prefect.resource.role"] = "worker"
related.append(RelatedResource(__root__=worker_resource))
return related
def _emit_flow_run_submitted_event(
self, configuration: BaseJobConfiguration
) -> Event:
return emit_event(
event="prefect.worker.submitted-flow-run",
resource=self._event_resource(),
related=self._event_related_resources(configuration=configuration),
)
def _emit_flow_run_executed_event(
self,
result: BaseWorkerResult,
configuration: BaseJobConfiguration,
submitted_event: Event,
):
related = self._event_related_resources(configuration=configuration)
for resource in related:
if resource.role == "flow-run":
resource["prefect.infrastructure.identifier"] = str(result.identifier)
resource["prefect.infrastructure.status-code"] = str(result.status_code)
emit_event(
event="prefect.worker.executed-flow-run",
resource=self._event_resource(),
related=related,
follows=submitted_event,
)
async def _emit_worker_started_event(self) -> Event:
return emit_event(
"prefect.worker.started",
resource=self._event_resource(),
related=self._event_related_resources(),
)
async def _emit_worker_stopped_event(self, started_event: Event):
emit_event(
"prefect.worker.stopped",
resource=self._event_resource(),
related=self._event_related_resources(),
follows=started_event,
)
def _emit_flow_run_cancelled_event(
self, flow_run: "FlowRun", configuration: BaseJobConfiguration
):
related = self._event_related_resources(configuration=configuration)
for resource in related:
if resource.role == "flow-run":
resource["prefect.infrastructure.identifier"] = str(
flow_run.infrastructure_pid
)
emit_event(
event="prefect.worker.cancelled-flow-run",
resource=self._event_resource(),
related=related,
)
|