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 | class JobRegistry(object):
def __init__(self, kiara: "Kiara"):
self._kiara: Kiara = kiara
self._job_matcher_cache: Dict[JobCacheStrategy, JobMatcher] = {}
self._active_jobs: bidict[str, uuid.UUID] = bidict()
self._failed_jobs: Dict[str, uuid.UUID] = {}
self._finished_jobs: Dict[str, uuid.UUID] = {}
self._archived_records: Dict[uuid.UUID, JobRecord] = {}
self._processor: ModuleProcessor = SynchronousProcessor(kiara=self._kiara)
self._processor.register_job_status_listener(self)
self._job_archives: Dict[str, JobArchive] = {}
self._default_job_store: Union[str, None] = None
self._event_callback = self._kiara.event_registry.add_producer(self)
# default_archive = FileSystemJobStore.create_from_kiara_context(self._kiara)
# self.register_job_archive(default_archive, store_alias=DEFAULT_STORE_MARKER)
# default_file_store = self._kiara.data_registry.get_archive(DEFAULT_STORE_MARKER)
# self.register_job_archive(default_file_store, store_alias="default_data_store") # type: ignore
@property
def job_matcher(self) -> JobMatcher:
from kiara.context.runtime_config import JobCacheStrategy
strategy = self._kiara.runtime_config.job_cache
if is_develop():
dev_config = get_dev_config()
if not dev_config.job_cache:
logger.debug(
"disable.job_cache",
reason="dev mode enabled and 'disable_job_cache' is set.",
)
strategy = JobCacheStrategy.no_cache
job_matcher = self._job_matcher_cache.get(strategy, None)
if job_matcher is None:
if strategy == JobCacheStrategy.no_cache:
job_matcher = NoneJobMatcher(kiara=self._kiara)
elif strategy == JobCacheStrategy.value_id:
job_matcher = ValueIdJobMatcher(kiara=self._kiara)
elif strategy == JobCacheStrategy.data_hash:
job_matcher = DataHashJobMatcher(kiara=self._kiara)
else:
raise Exception(f"Job cache strategy not implemented: {strategy}")
self._job_matcher_cache[strategy] = job_matcher
return job_matcher
def suppoerted_event_types(self) -> Iterable[Type[KiaraEvent]]:
return [JobArchiveAddedEvent, JobRecordPreStoreEvent, JobRecordStoredEvent]
def register_job_archive(self, archive: JobArchive, alias: Union[str, None] = None):
if alias is None:
alias = str(archive.archive_id)
if alias in self._job_archives.keys():
raise Exception(
f"Can't register job store, store id already registered: {alias}."
)
self._job_archives[alias] = archive
is_store = False
is_default_store = False
if isinstance(archive, JobStore):
is_store = True
if self._default_job_store is None:
self._default_job_store = alias
event = JobArchiveAddedEvent.construct(
kiara_id=self._kiara.id,
job_archive_id=archive.archive_id,
job_archive_alias=alias,
is_store=is_store,
is_default_store=is_default_store,
)
self._event_callback(event)
@property
def default_job_store(self) -> str:
if self._default_job_store is None:
raise Exception("No default job store set (yet).")
return self._default_job_store # type: ignore
def get_archive(self, store_id: Union[str, None] = None) -> JobArchive:
if store_id is None:
store_id = self.default_job_store
if store_id is None:
raise Exception("Can't retrieve deafult job archive, none set (yet).")
return self._job_archives[store_id]
@property
def job_archives(self) -> Mapping[str, JobArchive]:
return self._job_archives
def job_status_changed(
self,
job_id: uuid.UUID,
old_status: Union[JobStatus, None],
new_status: JobStatus,
):
# print(f"JOB STATUS CHANGED: {job_id} - {old_status} - {new_status.value}")
if job_id in self._active_jobs.values() and new_status is JobStatus.FAILED:
job_hash = self._active_jobs.inverse.pop(job_id)
self._failed_jobs[job_hash] = job_id
elif job_id in self._active_jobs.values() and new_status is JobStatus.SUCCESS:
job_hash = self._active_jobs.inverse.pop(job_id)
job_record = self._processor.get_job_record(job_id)
self._finished_jobs[job_hash] = job_id
self._archived_records[job_id] = job_record
def store_job_record(self, job_id: uuid.UUID):
if job_id not in self._archived_records.keys():
raise Exception(
f"Can't store job with id '{job_id}': no job record with that id exists."
)
job_record = self._archived_records[job_id]
if job_record._is_stored:
logger.debug(
"ignore.store.job_record", reason="already stored", job_id=str(job_id)
)
return
store: JobStore = self.get_archive() # type: ignore
if not isinstance(store, JobStore):
raise Exception("Can't store job record to archive: not writable.")
# if job_record.job_id in self._finished_jobs.values():
# logger.debug(
# "ignore.store.job_record",
# reason="already stored in store",
# job_id=str(job_id),
# )
# return
logger.debug(
"store.job_record",
job_hash=job_record.job_hash,
module_type=job_record.module_type,
)
pre_store_event = JobRecordPreStoreEvent.construct(
kiara_id=self._kiara.id, job_record=job_record
)
self._event_callback(pre_store_event)
store.store_job_record(job_record)
stored_event = JobRecordStoredEvent.construct(
kiara_id=self._kiara.id, job_record=job_record
)
self._event_callback(stored_event)
def get_job_record_in_session(self, job_id: uuid.UUID) -> JobRecord:
return self._processor.get_job_record(job_id)
def get_job_record(self, job_id: uuid.UUID) -> Union[JobRecord, None]:
if job_id in self._archived_records.keys():
return self._archived_records[job_id]
try:
job_record = self._processor.get_job_record(job_id=job_id)
return job_record
except Exception:
pass
job = self._processor.get_job(job_id=job_id)
if job is not None:
if job.status == JobStatus.FAILED:
return None
raise NotImplementedError()
def retrieve_all_job_records(self) -> Mapping[str, JobRecord]:
all_records: Dict[str, JobRecord] = {}
for archive in self.job_archives.values():
all_record_ids = archive.retrieve_all_job_hashes()
if all_record_ids is None:
return {}
for r in all_record_ids:
assert r not in all_records.keys()
job_record = archive.retrieve_record_for_job_hash(r)
assert job_record is not None
all_records[r] = job_record
return all_records
def find_matching_job_record(
self, inputs_manifest: InputsManifest
) -> Union[uuid.UUID, None]:
"""
Check if a job with same inputs manifest already ran some time before.
Arguments:
---------
inputs_manifest: the manifest incl. inputs
Returns:
-------
'None' if no such job exists, a (uuid) job-id if the job is currently running or has run in the past
"""
log = logger.bind(module_type=inputs_manifest.module_type)
if inputs_manifest.job_hash in self._active_jobs.keys():
log.debug("job.use_running")
return self._active_jobs[inputs_manifest.job_hash]
if inputs_manifest.job_hash in self._finished_jobs.keys():
job_id = self._finished_jobs[inputs_manifest.job_hash]
return job_id
module = self._kiara.module_registry.create_module(manifest=inputs_manifest)
if not module.characteristics.is_idempotent:
log.debug(
"skip.job_matching",
reason="module is not idempotent",
module_type=inputs_manifest.module_type,
)
return None
job_record = self.job_matcher.find_existing_job(inputs_manifest=inputs_manifest)
if job_record is None:
return None
self._finished_jobs[inputs_manifest.job_hash] = job_record.job_id
self._archived_records[job_record.job_id] = job_record
log.debug(
"job.found_cached_record",
job_id=str(job_record.job_id),
job_hash=inputs_manifest.job_hash,
module_type=inputs_manifest.module_type,
)
return job_record.job_id
def prepare_job_config(
self, manifest: Manifest, inputs: Mapping[str, Any]
) -> JobConfig:
module = self._kiara.module_registry.create_module(manifest=manifest)
job_config = JobConfig.create_from_module(
data_registry=self._kiara.data_registry, module=module, inputs=inputs
)
return job_config
def execute(
self,
manifest: Manifest,
inputs: Mapping[str, Any],
wait: bool = False,
job_metadata: Union[None, Any] = None,
) -> uuid.UUID:
job_config = self.prepare_job_config(manifest=manifest, inputs=inputs)
return self.execute_job(job_config, wait=wait, job_metadata=job_metadata)
def execute_job(
self,
job_config: JobConfig,
wait: bool = False,
job_metadata: Union[None, Any] = None,
) -> uuid.UUID:
log = logger.bind(
module_type=job_config.module_type,
module_config=job_config.module_config,
inputs={k: str(v) for k, v in job_config.inputs.items()},
job_hash=job_config.job_hash,
)
stored_job = self.find_matching_job_record(inputs_manifest=job_config)
if stored_job is not None:
log.debug(
"job.use_cached",
job_id=str(stored_job),
module_type=job_config.module_type,
)
return stored_job
if job_metadata is None:
job_metadata = {}
is_pipeline_step = job_metadata.get("is_pipeline_step", False)
dbg_data = {
"module_type": job_config.module_type,
"is_pipeline_step": is_pipeline_step,
}
if is_pipeline_step:
dbg_data["step_id"] = job_metadata["step_id"]
log.debug("job.execute", **dbg_data)
job_id = self._processor.create_job(
job_config=job_config, job_metadata=job_metadata
)
self._active_jobs[job_config.job_hash] = job_id
try:
self._processor.queue_job(job_id=job_id)
except Exception as e:
log.error("error.queue_job", job_id=job_id)
raise e
if wait:
self._processor.wait_for(job_id)
return job_id
def get_active_job(self, job_id: uuid.UUID) -> ActiveJob:
if job_id in self._active_jobs.keys() or job_id in self._failed_jobs.keys():
return self._processor.get_job(job_id)
else:
if job_id in self._archived_records.keys():
raise Exception(
f"Can't retrieve active job with id '{job_id}': job is archived."
)
elif job_id in self._processor._failed_jobs.keys():
job = self._processor.get_job(job_id)
msg = job.error
if not msg and job._exception:
msg = str(job._exception)
if not msg:
msg = repr(job._exception)
raise FailedJobException(job=job, msg=msg, parent=job._exception)
else:
raise Exception(f"Can't retrieve job with id '{job_id}': no such job.")
def get_job(self, job_id: uuid.UUID) -> ActiveJob:
return self._processor.get_job(job_id=job_id)
def get_job_status(self, job_id: uuid.UUID) -> JobStatus:
if job_id in self._archived_records.keys():
return JobStatus.SUCCESS
elif job_id in self._failed_jobs.values():
return JobStatus.FAILED
return self._processor.get_job_status(job_id=job_id)
def wait_for(self, *job_id: uuid.UUID):
not_finished = (j for j in job_id if j not in self._archived_records.keys())
if not_finished:
self._processor.wait_for(*not_finished)
def retrieve_result(self, job_id: uuid.UUID) -> ValueMapReadOnly:
if job_id not in self._archived_records.keys():
self._processor.wait_for(job_id)
if job_id in self._archived_records.keys():
job_record = self._archived_records[job_id]
results = self._kiara.data_registry.load_values(job_record.outputs)
return results
elif job_id in self._failed_jobs.values():
j = self._processor.get_job(job_id=job_id)
raise FailedJobException(job=j, parent=j._exception)
else:
raise Exception(f"Could not find job with id: {job_id}")
def execute_and_retrieve(
self, manifest: Manifest, inputs: Mapping[str, Any]
) -> ValueMap:
job_id = self.execute(manifest=manifest, inputs=inputs, wait=True)
results = self.retrieve_result(job_id=job_id)
return results
|