1 # Copyright (C) The Arvados Authors. All rights reserved.
3 # SPDX-License-Identifier: Apache-2.0
13 from functools import partial
18 from cwltool.errors import WorkflowException
19 import cwltool.workflow
20 from schema_salad.sourceline import SourceLine, cmap
21 import schema_salad.validate as validate
22 from schema_salad.ref_resolver import file_uri, uri_file_path
26 from arvados.keep import KeepClient
27 from arvados.errors import ApiError
29 import arvados_cwl.util
30 from .arvcontainer import RunnerContainer, cleanup_name_for_collection
31 from .runner import Runner, upload_docker, upload_job_order, upload_workflow_deps, make_builder, update_from_merged_map, print_keep_deps, ArvSecretStore
32 from .arvtool import ArvadosCommandTool, validate_cluster_target, ArvadosExpressionTool
33 from .arvworkflow import ArvadosWorkflow, upload_workflow, make_workflow_record
34 from .fsaccess import CollectionFsAccess, CollectionFetcher, collectionResolver, CollectionCache, pdh_size
35 from .perf import Perf
36 from .pathmapper import NoFollowPathMapper
37 from cwltool.task_queue import TaskQueue
38 from .context import ArvLoadingContext, ArvRuntimeContext
39 from ._version import __version__
41 from cwltool.process import shortname, UnsupportedRequirement, use_custom_schema
42 from cwltool.utils import adjustFileObjs, adjustDirObjs, get_listing, visit_class, aslist
43 from cwltool.command_line_tool import compute_checksums
44 from cwltool.load_tool import load_tool
46 logger = logging.getLogger('arvados.cwl-runner')
47 metrics = logging.getLogger('arvados.cwl-runner.metrics')
49 DEFAULT_PRIORITY = 500
51 class RuntimeStatusLoggingHandler(logging.Handler):
53 Intercepts logging calls and report them as runtime statuses on runner
56 def __init__(self, runtime_status_update_func):
57 super(RuntimeStatusLoggingHandler, self).__init__()
58 self.runtime_status_update = runtime_status_update_func
59 self.updatingRuntimeStatus = False
61 def emit(self, record):
63 if record.levelno >= logging.ERROR:
65 elif record.levelno >= logging.WARNING:
67 if kind == 'warning' and record.name in ("salad", "crunchstat_summary"):
68 # Don't send validation warnings to runtime status,
69 # they're noisy and unhelpful.
71 if kind is not None and self.updatingRuntimeStatus is not True:
72 self.updatingRuntimeStatus = True
74 log_msg = record.getMessage()
76 # If the logged message is multi-line, use its first line as status
77 # and the rest as detail.
78 status, detail = log_msg.split('\n', 1)
79 self.runtime_status_update(
81 "%s: %s" % (record.name, status),
85 self.runtime_status_update(
87 "%s: %s" % (record.name, record.getMessage())
90 self.updatingRuntimeStatus = False
93 class ArvCwlExecutor(object):
94 """Execute a CWL tool or workflow, submit work (using containers API),
95 wait for them to complete, and report output.
99 def __init__(self, api_client,
107 arvargs = argparse.Namespace()
108 arvargs.work_api = None
109 arvargs.output_name = None
110 arvargs.output_tags = None
111 arvargs.thread_count = 1
112 arvargs.collection_cache_size = None
113 arvargs.git_info = True
114 arvargs.submit = False
115 arvargs.defer_downloads = False
117 self.api = api_client
119 self.workflow_eval_lock = threading.Condition(threading.RLock())
120 self.final_output = None
121 self.final_status = None
122 self.num_retries = num_retries
124 self.stop_polling = threading.Event()
127 self.final_output_collection = None
128 self.output_name = arvargs.output_name
129 self.output_tags = arvargs.output_tags
130 self.project_uuid = None
131 self.intermediate_output_ttl = 0
132 self.intermediate_output_collections = []
133 self.trash_intermediate = False
134 self.thread_count = arvargs.thread_count
135 self.poll_interval = 12
136 self.loadingContext = None
137 self.should_estimate_cache_size = True
138 self.fs_access = None
139 self.secret_store = None
141 self.fast_submit = False
142 self.git_info = arvargs.git_info
145 if keep_client is not None:
146 self.keep_client = keep_client
148 self.keep_client = arvados.keep.KeepClient(api_client=self.api, num_retries=self.num_retries)
150 if arvargs.collection_cache_size:
151 collection_cache_size = arvargs.collection_cache_size*1024*1024
152 self.should_estimate_cache_size = False
154 collection_cache_size = 256*1024*1024
156 self.collection_cache = CollectionCache(self.api, self.keep_client, self.num_retries,
157 cap=collection_cache_size)
159 self.fetcher_constructor = partial(CollectionFetcher,
161 fs_access=CollectionFsAccess("", collection_cache=self.collection_cache),
162 num_retries=self.num_retries)
165 expected_api = ["containers"]
166 for api in expected_api:
168 methods = self.api._rootDesc.get('resources')[api]['methods']
169 if ('httpMethod' in methods['create'] and
170 (arvargs.work_api == api or arvargs.work_api is None)):
176 if not self.work_api:
177 if arvargs.work_api is None:
178 raise Exception("No supported APIs")
180 raise Exception("Unsupported API '%s', expected one of %s" % (arvargs.work_api, expected_api))
182 if self.work_api == "jobs":
184 *******************************
185 The 'jobs' API is no longer supported.
186 *******************************""")
189 self.loadingContext = ArvLoadingContext(vars(arvargs))
190 self.loadingContext.fetcher_constructor = self.fetcher_constructor
191 self.loadingContext.resolver = partial(collectionResolver, self.api, num_retries=self.num_retries)
192 self.loadingContext.construct_tool_object = self.arv_make_tool
194 # Add a custom logging handler to the root logger for runtime status reporting
195 # if running inside a container
196 if arvados_cwl.util.get_current_container(self.api, self.num_retries, logger):
197 root_logger = logging.getLogger('')
199 # Remove existing RuntimeStatusLoggingHandlers if they exist
200 handlers = [h for h in root_logger.handlers if not isinstance(h, RuntimeStatusLoggingHandler)]
201 root_logger.handlers = handlers
203 handler = RuntimeStatusLoggingHandler(self.runtime_status_update)
204 root_logger.addHandler(handler)
206 self.toplevel_runtimeContext = ArvRuntimeContext(vars(arvargs))
207 self.toplevel_runtimeContext.make_fs_access = partial(CollectionFsAccess,
208 collection_cache=self.collection_cache)
209 self.toplevel_runtimeContext.secret_store = ArvSecretStore()
211 self.defer_downloads = arvargs.submit and arvargs.defer_downloads
213 validate_cluster_target(self, self.toplevel_runtimeContext)
216 def arv_make_tool(self, toolpath_object, loadingContext):
217 if "class" in toolpath_object and toolpath_object["class"] == "CommandLineTool":
218 return ArvadosCommandTool(self, toolpath_object, loadingContext)
219 elif "class" in toolpath_object and toolpath_object["class"] == "Workflow":
220 return ArvadosWorkflow(self, toolpath_object, loadingContext)
221 elif "class" in toolpath_object and toolpath_object["class"] == "ExpressionTool":
222 return ArvadosExpressionTool(self, toolpath_object, loadingContext)
224 raise Exception("Unknown tool %s" % toolpath_object.get("class"))
226 def output_callback(self, out, processStatus):
227 with self.workflow_eval_lock:
228 if processStatus == "success":
229 logger.info("Overall process status is %s", processStatus)
232 logger.error("Overall process status is %s", processStatus)
235 self.api.pipeline_instances().update(uuid=self.pipeline["uuid"],
236 body={"state": state}).execute(num_retries=self.num_retries)
237 self.final_status = processStatus
238 self.final_output = out
239 self.workflow_eval_lock.notifyAll()
242 def start_run(self, runnable, runtimeContext):
243 self.task_queue.add(partial(runnable.run, runtimeContext),
244 self.workflow_eval_lock, self.stop_polling)
246 def process_submitted(self, container):
247 with self.workflow_eval_lock:
248 self.processes[container.uuid] = container
250 def process_done(self, uuid, record):
251 with self.workflow_eval_lock:
252 j = self.processes[uuid]
253 logger.info("%s %s is %s", self.label(j), uuid, record["state"])
254 self.task_queue.add(partial(j.done, record),
255 self.workflow_eval_lock, self.stop_polling)
256 del self.processes[uuid]
258 def runtime_status_update(self, kind, message, detail=None):
260 Updates the runtime_status field on the runner container.
261 Called when there's a need to report errors, warnings or just
262 activity statuses, for example in the RuntimeStatusLoggingHandler.
265 if kind not in ('error', 'warning', 'activity'):
266 # Ignore any other status kind
269 with self.workflow_eval_lock:
272 current = arvados_cwl.util.get_current_container(self.api, self.num_retries, logger)
273 except Exception as e:
274 logger.info("Couldn't get current container: %s", e)
277 runtime_status = current.get('runtime_status', {})
279 original_updatemessage = updatemessage = runtime_status.get(kind, "")
280 if kind == "activity" or not updatemessage:
281 updatemessage = message
283 # Subsequent messages tacked on in detail
284 original_updatedetail = updatedetail = runtime_status.get(kind+'Detail', "")
286 if updatedetail.count("\n") < maxlines:
289 updatedetail += message + "\n"
292 updatedetail += detail + "\n"
294 if updatedetail.count("\n") >= maxlines:
295 updatedetail += "\nSome messages may have been omitted. Check the full log."
297 if updatemessage == original_updatemessage and updatedetail == original_updatedetail:
298 # don't waste time doing an update if nothing changed
299 # (usually because we exceeded the max lines)
302 runtime_status.update({
304 kind+'Detail': updatedetail,
308 self.api.containers().update(uuid=current['uuid'],
310 'runtime_status': runtime_status,
311 }).execute(num_retries=self.num_retries)
312 except Exception as e:
313 logger.info("Couldn't update runtime_status: %s", e)
315 def wrapped_callback(self, cb, obj, st):
316 with self.workflow_eval_lock:
318 self.workflow_eval_lock.notifyAll()
320 def get_wrapped_callback(self, cb):
321 return partial(self.wrapped_callback, cb)
323 def on_message(self, event):
324 if event.get("object_uuid") in self.processes and event["event_type"] == "update":
325 uuid = event["object_uuid"]
326 if event["properties"]["new_attributes"]["state"] == "Running":
327 with self.workflow_eval_lock:
328 j = self.processes[uuid]
329 if j.running is False:
331 j.update_pipeline_component(event["properties"]["new_attributes"])
332 logger.info("%s %s is Running", self.label(j), uuid)
333 elif event["properties"]["new_attributes"]["state"] in ("Complete", "Failed", "Cancelled", "Final"):
334 self.process_done(uuid, event["properties"]["new_attributes"])
336 def label(self, obj):
337 return "[%s %s]" % (self.work_api[0:-1], obj.name)
339 def poll_states(self):
340 """Poll status of containers listed in the processes dict.
342 Runs in a separate thread.
346 remain_wait = self.poll_interval
349 self.stop_polling.wait(remain_wait)
350 if self.stop_polling.is_set():
352 with self.workflow_eval_lock:
353 keys = list(self.processes)
355 remain_wait = self.poll_interval
358 begin_poll = time.time()
359 if self.work_api == "containers":
360 table = self.poll_api.container_requests()
362 pageSize = self.poll_api._rootDesc.get('maxItemsPerResponse', 1000)
365 page = keys[:pageSize]
367 proc_states = table.list(filters=[["uuid", "in", page]], select=["uuid", "container_uuid", "state", "log_uuid",
368 "output_uuid", "modified_at", "properties",
369 "runtime_constraints"]).execute(num_retries=self.num_retries)
370 except Exception as e:
371 logger.warning("Temporary error checking states on API server: %s", e)
372 remain_wait = self.poll_interval
375 for p in proc_states["items"]:
377 "object_uuid": p["uuid"],
378 "event_type": "update",
383 keys = keys[pageSize:]
385 finish_poll = time.time()
386 remain_wait = self.poll_interval - (finish_poll - begin_poll)
388 logger.exception("Fatal error in state polling thread.")
389 with self.workflow_eval_lock:
390 self.processes.clear()
391 self.workflow_eval_lock.notifyAll()
393 self.stop_polling.set()
395 def add_intermediate_output(self, uuid):
397 self.intermediate_output_collections.append(uuid)
399 def trash_intermediate_output(self):
400 logger.info("Cleaning up intermediate output collections")
401 for i in self.intermediate_output_collections:
403 self.api.collections().delete(uuid=i).execute(num_retries=self.num_retries)
405 logger.warning("Failed to delete intermediate output: %s", sys.exc_info()[1], exc_info=(sys.exc_info()[1] if self.debug else False))
406 except (KeyboardInterrupt, SystemExit):
409 def check_features(self, obj, parentfield=""):
410 if isinstance(obj, dict):
411 if obj.get("class") == "DockerRequirement":
412 if obj.get("dockerOutputDirectory"):
413 if not obj.get("dockerOutputDirectory").startswith('/'):
414 raise SourceLine(obj, "dockerOutputDirectory", validate.ValidationException).makeError(
415 "Option 'dockerOutputDirectory' must be an absolute path.")
416 if obj.get("class") == "InplaceUpdateRequirement":
417 if obj["inplaceUpdate"] and parentfield == "requirements":
418 raise SourceLine(obj, "class", UnsupportedRequirement).makeError("InplaceUpdateRequirement not supported for keep collections.")
419 for k,v in obj.items():
420 self.check_features(v, parentfield=k)
421 elif isinstance(obj, list):
422 for i,v in enumerate(obj):
423 with SourceLine(obj, i, UnsupportedRequirement, logger.isEnabledFor(logging.DEBUG)):
424 self.check_features(v, parentfield=parentfield)
426 def make_output_collection(self, name, storage_classes, tagsString, output_properties, outputObj):
427 outputObj = copy.deepcopy(outputObj)
430 def capture(fileobj):
431 files.append(fileobj)
433 adjustDirObjs(outputObj, capture)
434 adjustFileObjs(outputObj, capture)
436 generatemapper = NoFollowPathMapper(files, "", "", separateDirs=False)
438 final = arvados.collection.Collection(api_client=self.api,
439 keep_client=self.keep_client,
440 num_retries=self.num_retries)
442 for k,v in generatemapper.items():
443 if v.type == "Directory" and v.resolved.startswith("_:"):
445 if v.type == "CreateFile" and (k.startswith("_:") or v.resolved.startswith("_:")):
446 with final.open(v.target, "wb") as f:
447 f.write(v.resolved.encode("utf-8"))
450 if not v.resolved.startswith("keep:"):
451 raise Exception("Output source is not in keep or a literal")
452 sp = v.resolved.split("/")
453 srccollection = sp[0][5:]
455 reader = self.collection_cache.get(srccollection)
456 srcpath = urllib.parse.unquote("/".join(sp[1:]) if len(sp) > 1 else ".")
457 final.copy(srcpath, v.target, source_collection=reader, overwrite=False)
458 except arvados.errors.ArgumentError as e:
459 logger.error("Creating CollectionReader for '%s' '%s': %s", k, v, e)
462 logger.error("While preparing output collection: %s", e)
465 def rewrite(fileobj):
466 fileobj["location"] = generatemapper.mapper(fileobj["location"]).target
467 for k in ("listing", "contents", "nameext", "nameroot", "dirname"):
471 adjustDirObjs(outputObj, rewrite)
472 adjustFileObjs(outputObj, rewrite)
474 with final.open("cwl.output.json", "w") as f:
475 res = str(json.dumps(outputObj, sort_keys=True, indent=4, separators=(',',': '), ensure_ascii=False))
479 final.save_new(name=name, owner_uuid=self.project_uuid, storage_classes=storage_classes,
480 ensure_unique_name=True, properties=output_properties)
482 logger.info("Final output collection %s \"%s\" (%s)", final.portable_data_hash(),
483 final.api_response()["name"],
484 final.manifest_locator())
486 final_uuid = final.manifest_locator()
487 tags = tagsString.split(',')
489 self.api.links().create(body={
490 "head_uuid": final_uuid, "link_class": "tag", "name": tag
491 }).execute(num_retries=self.num_retries)
493 def finalcollection(fileobj):
494 fileobj["location"] = "keep:%s/%s" % (final.portable_data_hash(), fileobj["location"])
496 adjustDirObjs(outputObj, finalcollection)
497 adjustFileObjs(outputObj, finalcollection)
499 return (outputObj, final)
501 def set_crunch_output(self):
502 if self.work_api == "containers":
503 current = arvados_cwl.util.get_current_container(self.api, self.num_retries, logger)
507 self.api.containers().update(uuid=current['uuid'],
509 'output': self.final_output_collection.portable_data_hash(),
510 'output_properties': self.final_output_collection.get_properties(),
511 }).execute(num_retries=self.num_retries)
512 self.api.collections().update(uuid=self.final_output_collection.manifest_locator(),
515 }).execute(num_retries=self.num_retries)
517 logger.exception("Setting container output")
520 def apply_reqs(self, job_order_object, tool):
521 if "https://w3id.org/cwl/cwl#requirements" in job_order_object:
522 if tool.metadata.get("http://commonwl.org/cwltool#original_cwlVersion") == 'v1.0':
523 raise WorkflowException(
524 "`cwl:requirements` in the input object is not part of CWL "
525 "v1.0. You can adjust to use `cwltool:overrides` instead; or you "
526 "can set the cwlVersion to v1.1 or greater and re-run with "
528 job_reqs = job_order_object["https://w3id.org/cwl/cwl#requirements"]
530 tool.requirements.append(req)
533 def get_git_info(tool):
534 in_a_git_repo = False
538 if tool.tool["id"].startswith("file://"):
539 # check if git is installed
541 filepath = uri_file_path(tool.tool["id"])
542 cwd = os.path.dirname(filepath)
544 ["git", "log", "--format=%H", "-n1", "HEAD"],
547 stdout=subprocess.DEVNULL,
550 except Exception as e:
557 return subprocess.run(
560 stdout=subprocess.PIPE,
561 universal_newlines=True,
563 git_commit = git_output(["git", "log", "--format=%H", "-n1", "HEAD"])
564 git_date = git_output(["git", "log", "--format=%cD", "-n1", "HEAD"])
565 git_committer = git_output(["git", "log", "--format=%cn <%ce>", "-n1", "HEAD"])
566 git_branch = git_output(["git", "rev-parse", "--abbrev-ref", "HEAD"])
567 git_origin = git_output(["git", "remote", "get-url", "origin"])
568 git_status = git_output(["git", "status", "--untracked-files=no", "--porcelain"])
569 git_describe = git_output(["git", "describe", "--always", "--tags"])
570 git_toplevel = git_output(["git", "rev-parse", "--show-toplevel"])
571 git_path = filepath[len(git_toplevel):]
574 "http://arvados.org/cwl#gitCommit": git_commit,
575 "http://arvados.org/cwl#gitDate": git_date,
576 "http://arvados.org/cwl#gitCommitter": git_committer,
577 "http://arvados.org/cwl#gitBranch": git_branch,
578 "http://arvados.org/cwl#gitOrigin": git_origin,
579 "http://arvados.org/cwl#gitStatus": git_status,
580 "http://arvados.org/cwl#gitDescribe": git_describe,
581 "http://arvados.org/cwl#gitPath": git_path,
584 for g in ("http://arvados.org/cwl#gitCommit",
585 "http://arvados.org/cwl#gitDate",
586 "http://arvados.org/cwl#gitCommitter",
587 "http://arvados.org/cwl#gitBranch",
588 "http://arvados.org/cwl#gitOrigin",
589 "http://arvados.org/cwl#gitStatus",
590 "http://arvados.org/cwl#gitDescribe",
591 "http://arvados.org/cwl#gitPath"):
592 if g in tool.metadata:
593 gitproperties[g] = tool.metadata[g]
597 def set_container_request_properties(self, container, properties):
598 resp = self.api.container_requests().list(filters=[["container_uuid", "=", container["uuid"]]], select=["uuid", "properties"]).execute(num_retries=self.num_retries)
599 for cr in resp["items"]:
600 cr["properties"].update({k.replace("http://arvados.org/cwl#", "arv:"): v for k, v in properties.items()})
601 self.api.container_requests().update(uuid=cr["uuid"], body={"container_request": {"properties": cr["properties"]}}).execute(num_retries=self.num_retries)
603 def arv_executor(self, updated_tool, job_order, runtimeContext, logger=None):
604 self.debug = runtimeContext.debug
606 self.runtime_status_update("activity", "initialization")
608 git_info = self.get_git_info(updated_tool) if self.git_info else {}
610 logger.info("Git provenance")
613 logger.info(" %s: %s", g.split("#", 1)[1], git_info[g])
615 runtimeContext.git_info = git_info
617 workbench1 = self.api.config()["Services"]["Workbench1"]["ExternalURL"]
618 workbench2 = self.api.config()["Services"]["Workbench2"]["ExternalURL"]
619 controller = self.api.config()["Services"]["Controller"]["ExternalURL"]
620 logger.info("Using cluster %s (%s)", self.api.config()["ClusterID"], workbench2 or workbench1 or controller)
622 if not self.fast_submit:
623 updated_tool.visit(self.check_features)
626 self.fs_access = runtimeContext.make_fs_access(runtimeContext.basedir)
627 self.secret_store = runtimeContext.secret_store
629 self.trash_intermediate = runtimeContext.trash_intermediate
630 if self.trash_intermediate and self.work_api != "containers":
631 raise Exception("--trash-intermediate is only supported with --api=containers.")
633 self.intermediate_output_ttl = runtimeContext.intermediate_output_ttl
634 if self.intermediate_output_ttl and self.work_api != "containers":
635 raise Exception("--intermediate-output-ttl is only supported with --api=containers.")
636 if self.intermediate_output_ttl < 0:
637 raise Exception("Invalid value %d for --intermediate-output-ttl, cannot be less than zero" % self.intermediate_output_ttl)
639 if runtimeContext.submit_request_uuid and self.work_api != "containers":
640 raise Exception("--submit-request-uuid requires containers API, but using '{}' api".format(self.work_api))
642 runtimeContext = runtimeContext.copy()
644 default_storage_classes = ",".join([k for k,v in self.api.config().get("StorageClasses", {"default": {"Default": True}}).items() if v.get("Default") is True])
645 if runtimeContext.storage_classes == "default":
646 runtimeContext.storage_classes = default_storage_classes
647 if runtimeContext.intermediate_storage_classes == "default":
648 runtimeContext.intermediate_storage_classes = default_storage_classes
650 if not runtimeContext.name:
651 self.name = updated_tool.tool.get("label") or updated_tool.metadata.get("label") or os.path.basename(updated_tool.tool["id"])
652 if git_info.get("http://arvados.org/cwl#gitDescribe"):
653 self.name = "%s (%s)" % (self.name, git_info.get("http://arvados.org/cwl#gitDescribe"))
654 runtimeContext.name = self.name
656 if runtimeContext.copy_deps is None and (runtimeContext.create_workflow or runtimeContext.update_workflow):
657 # When creating or updating workflow record, by default
658 # always copy dependencies and ensure Docker images are up
660 runtimeContext.copy_deps = True
661 runtimeContext.match_local_docker = True
663 if runtimeContext.print_keep_deps:
664 runtimeContext.copy_deps = False
665 runtimeContext.match_local_docker = False
667 if runtimeContext.update_workflow and self.project_uuid is None:
668 # If we are updating a workflow, make sure anything that
669 # gets uploaded goes into the same parent project, unless
670 # an alternate --project-uuid was provided.
671 existing_wf = self.api.workflows().get(uuid=runtimeContext.update_workflow).execute()
672 runtimeContext.project_uuid = existing_wf["owner_uuid"]
674 self.project_uuid = runtimeContext.project_uuid
676 self.runtime_status_update("activity", "data transfer")
678 # Upload local file references in the job order.
679 with Perf(metrics, "upload_job_order"):
680 job_order, jobmapper = upload_job_order(self, "%s input" % runtimeContext.name,
681 updated_tool, job_order, runtimeContext)
683 # determine if we are submitting or directly executing the workflow.
685 # the last clause means: if it is a command line tool, and we
686 # are going to wait for the result, and always_submit_runner
687 # is false, then we don't submit a runner process.
689 submitting = (runtimeContext.submit and not
690 (updated_tool.tool["class"] == "CommandLineTool" and
691 runtimeContext.wait and
692 not runtimeContext.always_submit_runner))
694 loadingContext = self.loadingContext.copy()
695 loadingContext.do_validate = False
696 loadingContext.disable_js_validation = True
699 # Upload direct dependencies of workflow steps, get back mapping of files to keep references.
700 # Also uploads docker images.
701 if not self.fast_submit:
702 logger.info("Uploading workflow dependencies")
703 with Perf(metrics, "upload_workflow_deps"):
704 merged_map = upload_workflow_deps(self, tool, runtimeContext)
706 # in the fast submit case, we are running a workflow that
707 # has already been uploaded to Arvados, so we assume all
708 # the dependencies have been pinned to keep references and
709 # there is nothing to do.
712 loadingContext.loader = tool.doc_loader
713 loadingContext.avsc_names = tool.doc_schema
714 loadingContext.metadata = tool.metadata
715 loadingContext.skip_resolve_all = True
717 workflow_wrapper = None
718 if (submitting and not self.fast_submit) or runtimeContext.update_workflow or runtimeContext.create_workflow or runtimeContext.print_keep_deps:
719 # upload workflow and get back the workflow wrapper
721 workflow_wrapper = upload_workflow(self, tool, job_order,
722 runtimeContext.project_uuid,
724 uuid=runtimeContext.update_workflow,
725 submit_runner_ram=runtimeContext.submit_runner_ram,
726 name=runtimeContext.name,
727 merged_map=merged_map,
728 submit_runner_image=runtimeContext.submit_runner_image,
730 set_defaults=(runtimeContext.update_workflow or runtimeContext.create_workflow),
733 if runtimeContext.update_workflow or runtimeContext.create_workflow:
734 # We're registering the workflow, so create or update
735 # the workflow record and then exit.
736 uuid = make_workflow_record(self, workflow_wrapper, runtimeContext.name, tool,
737 runtimeContext.project_uuid, runtimeContext.update_workflow)
738 self.stdout.write(uuid + "\n")
739 return (None, "success")
741 if runtimeContext.print_keep_deps:
742 # Just find and print out all the collection dependencies and exit
743 print_keep_deps(self, runtimeContext, merged_map, tool)
744 return (None, "success")
746 # Did not register a workflow, we're going to submit
748 loadingContext.loader.idx.clear()
749 loadingContext.loader.idx["_:main"] = workflow_wrapper
750 workflow_wrapper["id"] = "_:main"
752 # Reload the minimal wrapper workflow.
753 self.fast_submit = True
754 tool = load_tool(workflow_wrapper, loadingContext)
755 loadingContext.loader.idx["_:main"] = workflow_wrapper
758 # If we are going to run the workflow now (rather than
759 # submit it), we need to update the workflow document
760 # replacing file references with keep references. If we
761 # are just going to construct a run submission, we don't
763 update_from_merged_map(tool, merged_map)
765 self.apply_reqs(job_order, tool)
767 self.ignore_docker_for_reuse = runtimeContext.ignore_docker_for_reuse
768 self.eval_timeout = runtimeContext.eval_timeout
770 runtimeContext.use_container = True
771 runtimeContext.tmpdir_prefix = "tmp"
772 runtimeContext.work_api = self.work_api
774 if not self.output_name:
775 self.output_name = "Output from workflow %s" % runtimeContext.name
777 self.output_name = cleanup_name_for_collection(self.output_name)
779 if self.work_api == "containers":
780 if self.ignore_docker_for_reuse:
781 raise Exception("--ignore-docker-for-reuse not supported with containers API.")
782 runtimeContext.outdir = "/var/spool/cwl"
783 runtimeContext.docker_outdir = "/var/spool/cwl"
784 runtimeContext.tmpdir = "/tmp"
785 runtimeContext.docker_tmpdir = "/tmp"
787 if runtimeContext.priority < 1 or runtimeContext.priority > 1000:
788 raise Exception("--priority must be in the range 1..1000.")
790 if self.should_estimate_cache_size:
793 def estimate_collection_cache(obj):
794 if obj.get("location", "").startswith("keep:"):
795 m = pdh_size.match(obj["location"][5:])
796 if m and m.group(1) not in visited:
797 visited.add(m.group(1))
798 estimated_size[0] += int(m.group(2))
799 visit_class(job_order, ("File", "Directory"), estimate_collection_cache)
800 runtimeContext.collection_cache_size = max(((estimated_size[0]*192) // (1024*1024))+1, 256)
801 self.collection_cache.set_cap(runtimeContext.collection_cache_size*1024*1024)
803 logger.info("Using collection cache size %s MiB", runtimeContext.collection_cache_size)
806 if runtimeContext.submit:
807 # We are submitting instead of running immediately.
809 # Create a "Runner job" that when run() is invoked,
810 # creates the container request to run the workflow.
811 if self.work_api == "containers":
813 loadingContext.metadata = updated_tool.metadata.copy()
814 tool = RunnerContainer(self, tool, loadingContext, runtimeContext.enable_reuse,
817 submit_runner_ram=runtimeContext.submit_runner_ram,
818 name=runtimeContext.name,
819 on_error=runtimeContext.on_error,
820 submit_runner_image=runtimeContext.submit_runner_image,
821 intermediate_output_ttl=runtimeContext.intermediate_output_ttl,
822 merged_map=merged_map,
823 priority=runtimeContext.priority,
824 secret_store=self.secret_store,
825 collection_cache_size=runtimeContext.collection_cache_size,
826 collection_cache_is_default=self.should_estimate_cache_size,
829 runtimeContext.runnerjob = tool.tool["id"]
831 if runtimeContext.cwl_runner_job is not None:
832 self.uuid = runtimeContext.cwl_runner_job.get('uuid')
834 jobiter = tool.job(job_order,
835 self.output_callback,
838 if runtimeContext.submit and not runtimeContext.wait:
839 # User provided --no-wait so submit the container request,
840 # get the container request uuid, print it out, and exit.
841 runnerjob = next(jobiter)
842 runnerjob.run(runtimeContext)
843 self.stdout.write(runnerjob.uuid+"\n")
844 return (None, "success")
846 # We either running the workflow directly, or submitting it
847 # and will wait for a final result.
849 self.runtime_status_update("activity", "workflow execution")
851 current_container = arvados_cwl.util.get_current_container(self.api, self.num_retries, logger)
852 if current_container:
853 logger.info("Running inside container %s", current_container.get("uuid"))
854 self.set_container_request_properties(current_container, git_info)
856 self.poll_api = arvados.api('v1', timeout=runtimeContext.http_timeout)
857 self.polling_thread = threading.Thread(target=self.poll_states)
858 self.polling_thread.start()
860 self.task_queue = TaskQueue(self.workflow_eval_lock, self.thread_count)
863 self.workflow_eval_lock.acquire()
865 # Holds the lock while this code runs and releases it when
866 # it is safe to do so in self.workflow_eval_lock.wait(),
867 # at which point on_message can update job state and
868 # process output callbacks.
870 loopperf = Perf(metrics, "jobiter")
872 for runnable in jobiter:
875 if self.stop_polling.is_set():
878 if self.task_queue.error is not None:
879 raise self.task_queue.error
882 with Perf(metrics, "run"):
883 self.start_run(runnable, runtimeContext)
885 if (self.task_queue.in_flight + len(self.processes)) > 0:
886 self.workflow_eval_lock.wait(3)
888 if self.final_status is None:
889 logger.error("Workflow is deadlocked, no runnable processes and not waiting on any pending processes.")
892 if self.stop_polling.is_set():
898 while (self.task_queue.in_flight + len(self.processes)) > 0:
899 if self.task_queue.error is not None:
900 raise self.task_queue.error
901 self.workflow_eval_lock.wait(3)
903 except UnsupportedRequirement:
906 if sys.exc_info()[0] is KeyboardInterrupt or sys.exc_info()[0] is SystemExit:
907 logger.error("Interrupted, workflow will be cancelled")
908 elif isinstance(sys.exc_info()[1], WorkflowException):
909 logger.error("Workflow execution failed:\n%s", sys.exc_info()[1], exc_info=(sys.exc_info()[1] if self.debug else False))
911 logger.exception("Workflow execution failed")
914 self.api.pipeline_instances().update(uuid=self.pipeline["uuid"],
915 body={"state": "Failed"}).execute(num_retries=self.num_retries)
917 if self.work_api == "containers" and not current_container:
918 # Not running in a crunch container, so cancel any outstanding processes.
919 for p in self.processes:
921 self.api.container_requests().update(uuid=p,
922 body={"priority": "0"}
923 ).execute(num_retries=self.num_retries)
927 self.workflow_eval_lock.release()
928 self.task_queue.drain()
929 self.stop_polling.set()
930 self.polling_thread.join()
931 self.task_queue.join()
933 if self.final_status == "UnsupportedRequirement":
934 raise UnsupportedRequirement("Check log for details.")
936 if self.final_output is None:
937 raise WorkflowException("Workflow did not return a result.")
939 if runtimeContext.usage_report_notes:
940 logger.info("Steps with low resource utilization (possible optimization opportunities):")
941 for x in runtimeContext.usage_report_notes:
942 logger.info(" %s", x)
944 if runtimeContext.submit and isinstance(tool, Runner):
945 logger.info("Final output collection %s", tool.final_output)
946 if workbench2 or workbench1:
947 logger.info("Output at %scollections/%s", workbench2 or workbench1, tool.final_output)
949 if self.output_tags is None:
950 self.output_tags = ""
953 storage_class_req, _ = tool.get_requirement("http://arvados.org/cwl#OutputStorageClass")
954 if storage_class_req and storage_class_req.get("finalStorageClass"):
955 storage_classes = aslist(storage_class_req["finalStorageClass"])
957 storage_classes = runtimeContext.storage_classes.strip().split(",")
959 output_properties = {}
960 output_properties_req, _ = tool.get_requirement("http://arvados.org/cwl#OutputCollectionProperties")
961 if output_properties_req:
962 builder = make_builder(job_order, tool.hints, tool.requirements, runtimeContext, tool.metadata)
963 for pr in output_properties_req["outputProperties"]:
964 output_properties[pr["propertyName"]] = builder.do_eval(pr["propertyValue"])
966 self.final_output, self.final_output_collection = self.make_output_collection(self.output_name, storage_classes,
967 self.output_tags, output_properties,
969 self.set_crunch_output()
971 if runtimeContext.compute_checksum:
972 adjustDirObjs(self.final_output, partial(get_listing, self.fs_access))
973 adjustFileObjs(self.final_output, partial(compute_checksums, self.fs_access))
975 if self.trash_intermediate and self.final_status == "success":
976 self.trash_intermediate_output()
978 return (self.final_output, self.final_status)
980 def blank_secrets(job_order_object, process):
981 secrets_req, _ = process.get_requirement("http://commonwl.org/cwltool#Secrets")