1 # Copyright (C) The Arvados Authors. All rights reserved.
3 # SPDX-License-Identifier: Apache-2.0
5 from future import standard_library
6 standard_library.install_aliases()
7 from builtins import str
8 from builtins import next
22 import arvados_cwl.executor
23 import cwltool.process
24 from arvados.errors import ApiError
25 from schema_salad.ref_resolver import Loader
26 from schema_salad.sourceline import cmap
27 from .mock_discovery import get_rootDesc
28 from .matcher import JsonDiffMatcher, StripYAMLComments
29 from .test_container import CollectionMock
30 from arvados_cwl.arvdocker import arv_docker_clear_cache
32 if not os.getenv('ARVADOS_DEBUG'):
33 logging.getLogger('arvados.cwl-runner').setLevel(logging.WARN)
34 logging.getLogger('arvados.arv-run').setLevel(logging.WARN)
36 class TestJob(unittest.TestCase):
39 cwltool.process._names = set()
41 def helper(self, runner, enable_reuse=True):
42 document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.1")
44 make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
45 collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
46 loadingContext = arvados_cwl.context.ArvLoadingContext(
47 {"avsc_names": avsc_names,
49 "make_fs_access": make_fs_access,
51 "metadata": {"cwlVersion": "v1.1", "http://commonwl.org/cwltool#original_cwlVersion": "v1.0"},
52 "makeTool": runner.arv_make_tool})
53 runtimeContext = arvados_cwl.context.ArvRuntimeContext(
56 "name": "test_run_job_"+str(enable_reuse),
57 "make_fs_access": make_fs_access,
58 "enable_reuse": enable_reuse,
61 return loadingContext, runtimeContext
63 # The test passes no builder.resources
64 # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
65 @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
66 def test_run(self, list_images_in_arv):
67 for enable_reuse in (True, False):
68 arv_docker_clear_cache()
69 runner = mock.MagicMock()
70 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
71 runner.ignore_docker_for_reuse = False
72 runner.num_retries = 0
74 list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
75 runner.api.collections().get().execute.return_value = {"portable_data_hash": "99999999999999999999999999999993+99"}
76 # Simulate reused job from another project so that we can check is a can_read
78 runner.api.jobs().create().execute.return_value = {
79 'state': 'Complete' if enable_reuse else 'Queued',
80 'owner_uuid': 'zzzzz-tpzed-yyyyyyyyyyyyyyy' if enable_reuse else 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
81 'uuid': 'zzzzz-819sb-yyyyyyyyyyyyyyy',
89 "arguments": [{"valueFrom": "$(runtime.outdir)"}],
91 "class": "CommandLineTool"
94 loadingContext, runtimeContext = self.helper(runner, enable_reuse)
96 arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, loadingContext)
97 arvtool.formatgraph = None
98 for j in arvtool.job({}, mock.MagicMock(), runtimeContext):
100 runner.api.jobs().create.assert_called_with(
101 body=JsonDiffMatcher({
102 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
103 'runtime_constraints': {},
104 'script_parameters': {
106 'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
107 'command': ['ls', '$(task.outdir)']
110 'script_version': 'master',
111 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
112 'repository': 'arvados',
113 'script': 'crunchrunner',
114 'runtime_constraints': {
115 'docker_image': 'arvados/jobs',
116 'min_cores_per_node': 1,
117 'min_ram_mb_per_node': 1024,
118 'min_scratch_mb_per_node': 2048 # tmpdirSize + outdirSize
121 find_or_create=enable_reuse,
122 filters=[['repository', '=', 'arvados'],
123 ['script', '=', 'crunchrunner'],
124 ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
125 ['docker_image_locator', 'in docker', 'arvados/jobs']]
128 runner.api.links().create.assert_called_with(
129 body=JsonDiffMatcher({
130 'link_class': 'permission',
132 "tail_uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz",
133 "head_uuid": "zzzzz-819sb-yyyyyyyyyyyyyyy",
136 # Simulate an API excepction when trying to create a
137 # sharing link on the job
138 runner.api.links().create.side_effect = ApiError(
139 mock.MagicMock(return_value={'status': 403}),
140 bytes(b'Permission denied'))
141 j.run(runtimeContext)
143 assert not runner.api.links().create.called
145 # The test passes some fields in builder.resources
146 # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
147 @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
148 def test_resource_requirements(self, list_images_in_arv):
149 runner = mock.MagicMock()
150 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
151 runner.ignore_docker_for_reuse = False
152 runner.num_retries = 0
153 arvados_cwl.add_arv_hints()
155 list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
156 runner.api.collections().get().execute.return_value = {"portable_data_hash": "99999999999999999999999999999993+99"}
162 "class": "ResourceRequirement",
167 "class": "http://arvados.org/cwl#RuntimeConstraints",
169 "outputDirType": "keep_output_dir"
171 "class": "http://arvados.org/cwl#APIRequirement",
174 "class": "http://arvados.org/cwl#ReuseRequirement",
179 "class": "CommandLineTool"
182 loadingContext, runtimeContext = self.helper(runner)
184 arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, loadingContext)
185 arvtool.formatgraph = None
186 for j in arvtool.job({}, mock.MagicMock(), runtimeContext):
187 j.run(runtimeContext)
188 runner.api.jobs().create.assert_called_with(
189 body=JsonDiffMatcher({
190 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
191 'runtime_constraints': {},
192 'script_parameters': {
194 'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
195 'task.keepTmpOutput': True,
199 'script_version': 'master',
200 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
201 'repository': 'arvados',
202 'script': 'crunchrunner',
203 'runtime_constraints': {
204 'docker_image': 'arvados/jobs',
205 'min_cores_per_node': 3,
206 'min_ram_mb_per_node': 3512, # ramMin + keep_cache
207 'min_scratch_mb_per_node': 5024, # tmpdirSize + outdirSize
208 'keep_cache_mb_per_task': 512
211 find_or_create=False,
212 filters=[['repository', '=', 'arvados'],
213 ['script', '=', 'crunchrunner'],
214 ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
215 ['docker_image_locator', 'in docker', 'arvados/jobs']])
217 @mock.patch("arvados.collection.CollectionReader")
218 def test_done(self, reader):
219 api = mock.MagicMock()
221 runner = mock.MagicMock()
223 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
224 runner.num_retries = 0
225 runner.ignore_docker_for_reuse = False
227 reader().keys.return_value = "log.txt"
228 reader().open.return_value = io.StringIO(
229 str(u"""2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.tmpdir)=/tmp/crunch-job-task-work/compute3.1/tmpdir
230 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.outdir)=/tmp/crunch-job-task-work/compute3.1/outdir
231 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
233 api.collections().list().execute.side_effect = ({"items": []},
234 {"items": [{"manifest_text": "XYZ"}]},
236 {"items": [{"manifest_text": "ABC"}]})
238 arvjob = arvados_cwl.ArvadosJob(runner,
245 arvjob.output_callback = mock.MagicMock()
246 arvjob.collect_outputs = mock.MagicMock()
247 arvjob.collect_outputs.return_value = {"out": "stuff"}
251 "output": "99999999999999999999999999999993+99",
252 "log": "99999999999999999999999999999994+99",
253 "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
256 api.collections().list.assert_has_calls([
258 # Output collection check
259 mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
260 ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
261 ['name', '=', 'Output 9999999 of testjob']]),
262 mock.call().execute(num_retries=0),
263 mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999993+99']],
264 select=['manifest_text']),
265 mock.call().execute(num_retries=0),
266 # Log collection's turn
267 mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
268 ['portable_data_hash', '=', '99999999999999999999999999999994+99'],
269 ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]),
270 mock.call().execute(num_retries=0),
271 mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999994+99']],
272 select=['manifest_text']),
273 mock.call().execute(num_retries=0)])
275 api.collections().create.assert_has_calls([
276 mock.call(ensure_unique_name=True,
277 body={'portable_data_hash': '99999999999999999999999999999993+99',
278 'manifest_text': 'XYZ',
279 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
280 'name': 'Output 9999999 of testjob'}),
281 mock.call().execute(num_retries=0),
282 mock.call(ensure_unique_name=True,
283 body={'portable_data_hash': '99999999999999999999999999999994+99',
284 'manifest_text': 'ABC',
285 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
286 'name': 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
287 mock.call().execute(num_retries=0),
290 arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
292 @mock.patch("arvados.collection.CollectionReader")
293 def test_done_use_existing_collection(self, reader):
294 api = mock.MagicMock()
296 runner = mock.MagicMock()
298 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
299 runner.num_retries = 0
301 reader().keys.return_value = "log.txt"
302 reader().open.return_value = io.StringIO(
303 str(u"""2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.tmpdir)=/tmp/crunch-job-task-work/compute3.1/tmpdir
304 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.outdir)=/tmp/crunch-job-task-work/compute3.1/outdir
305 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
308 api.collections().list().execute.side_effect = (
309 {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},
310 {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},
313 arvjob = arvados_cwl.ArvadosJob(runner,
320 arvjob.output_callback = mock.MagicMock()
321 arvjob.collect_outputs = mock.MagicMock()
322 arvjob.collect_outputs.return_value = {"out": "stuff"}
326 "output": "99999999999999999999999999999993+99",
327 "log": "99999999999999999999999999999994+99",
328 "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
331 api.collections().list.assert_has_calls([
334 mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
335 ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
336 ['name', '=', 'Output 9999999 of testjob']]),
337 mock.call().execute(num_retries=0),
339 mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
340 ['portable_data_hash', '=', '99999999999999999999999999999994+99'],
341 ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]),
342 mock.call().execute(num_retries=0)
345 self.assertFalse(api.collections().create.called)
347 arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
350 class TestWorkflow(unittest.TestCase):
353 cwltool.process._names = set()
355 def helper(self, runner, enable_reuse=True):
356 document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.1")
358 make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
359 collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
361 document_loader.fetcher_constructor = functools.partial(arvados_cwl.CollectionFetcher, api_client=runner.api, fs_access=make_fs_access(""))
362 document_loader.fetcher = document_loader.fetcher_constructor(document_loader.cache, document_loader.session)
363 document_loader.fetch_text = document_loader.fetcher.fetch_text
364 document_loader.check_exists = document_loader.fetcher.check_exists
366 loadingContext = arvados_cwl.context.ArvLoadingContext(
367 {"avsc_names": avsc_names,
369 "make_fs_access": make_fs_access,
370 "loader": document_loader,
371 "metadata": {"cwlVersion": "v1.1", "http://commonwl.org/cwltool#original_cwlVersion": "v1.0"},
372 "construct_tool_object": runner.arv_make_tool})
373 runtimeContext = arvados_cwl.context.ArvRuntimeContext(
376 "name": "test_run_wf_"+str(enable_reuse),
377 "make_fs_access": make_fs_access,
378 "enable_reuse": enable_reuse,
381 return loadingContext, runtimeContext
383 # The test passes no builder.resources
384 # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
385 @mock.patch("arvados.collection.CollectionReader")
386 @mock.patch("arvados.collection.Collection")
387 @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
388 def test_run(self, list_images_in_arv, mockcollection, mockcollectionreader):
389 arv_docker_clear_cache()
390 arvados_cwl.add_arv_hints()
392 api = mock.MagicMock()
393 api._rootDesc = get_rootDesc()
395 runner = arvados_cwl.executor.ArvCwlExecutor(api, argparse.Namespace(work_api="jobs",
399 collection_cache_size=None))
400 self.assertEqual(runner.work_api, 'jobs')
402 list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
403 runner.api.collections().get().execute.return_value = {"portable_data_hash": "99999999999999999999999999999993+99"}
404 runner.api.collections().list().execute.return_value = {"items": [{
405 "uuid": "zzzzz-4zz18-zzzzzzzzzzzzzzz",
406 "portable_data_hash": "99999999999999999999999999999993+99"}]}
408 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
409 runner.ignore_docker_for_reuse = False
410 runner.num_retries = 0
412 loadingContext, runtimeContext = self.helper(runner)
413 runner.fs_access = runtimeContext.make_fs_access(runtimeContext.basedir)
414 tool, metadata = loadingContext.loader.resolve_ref("tests/wf/scatter2.cwl")
415 metadata["cwlVersion"] = tool["cwlVersion"]
417 mockc = mock.MagicMock()
418 mockcollection.side_effect = lambda *args, **kwargs: CollectionMock(mockc, *args, **kwargs)
419 mockcollectionreader().find.return_value = arvados.arvfile.ArvadosFile(mock.MagicMock(), "token.txt")
421 arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, loadingContext)
422 arvtool.formatgraph = None
423 it = arvtool.job({}, mock.MagicMock(), runtimeContext)
425 next(it).run(runtimeContext)
426 next(it).run(runtimeContext)
428 with open("tests/wf/scatter2_subwf.cwl") as f:
429 subwf = StripYAMLComments(f.read().rstrip())
431 runner.api.jobs().create.assert_called_with(
432 body=JsonDiffMatcher({
433 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
434 'repository': 'arvados',
435 'script_version': 'master',
436 'script': 'crunchrunner',
437 'script_parameters': {
438 'tasks': [{'task.env': {
439 'HOME': '$(task.outdir)',
440 'TMPDIR': '$(task.tmpdir)'},
442 'workflow.cwl': '$(task.keep)/99999999999999999999999999999996+99/workflow.cwl',
443 'cwl.input.yml': '$(task.keep)/99999999999999999999999999999996+99/cwl.input.yml'
445 'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'],
446 'task.stdout': 'cwl.output.json'}]},
447 'runtime_constraints': {
448 'min_scratch_mb_per_node': 2048,
449 'min_cores_per_node': 1,
450 'docker_image': 'arvados/jobs',
451 'min_ram_mb_per_node': 1024
453 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
454 filters=[['repository', '=', 'arvados'],
455 ['script', '=', 'crunchrunner'],
456 ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
457 ['docker_image_locator', 'in docker', 'arvados/jobs']],
460 mockc.open().__enter__().write.assert_has_calls([mock.call(subwf)])
461 mockc.open().__enter__().write.assert_has_calls([mock.call(
464 "basename": "token.txt",
466 "location": "/keep/99999999999999999999999999999999+118/token.txt",
472 # The test passes no builder.resources
473 # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
474 @mock.patch("arvados.collection.CollectionReader")
475 @mock.patch("arvados.collection.Collection")
476 @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
477 def test_overall_resource_singlecontainer(self, list_images_in_arv, mockcollection, mockcollectionreader):
478 arv_docker_clear_cache()
479 arvados_cwl.add_arv_hints()
481 api = mock.MagicMock()
482 api._rootDesc = get_rootDesc()
484 runner = arvados_cwl.executor.ArvCwlExecutor(api, argparse.Namespace(work_api="jobs",
488 collection_cache_size=None))
489 self.assertEqual(runner.work_api, 'jobs')
491 list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
492 runner.api.collections().get().execute.return_value = {"portable_data_hash": "99999999999999999999999999999993+99"}
493 runner.api.collections().list().execute.return_value = {"items": [{
494 "uuid": "zzzzz-4zz18-zzzzzzzzzzzzzzz",
495 "portable_data_hash": "99999999999999999999999999999993+99"}]}
497 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
498 runner.ignore_docker_for_reuse = False
499 runner.num_retries = 0
501 loadingContext, runtimeContext = self.helper(runner)
502 loadingContext.do_update = True
503 runner.fs_access = runtimeContext.make_fs_access(runtimeContext.basedir)
504 tool, metadata = loadingContext.loader.resolve_ref("tests/wf/echo-wf.cwl")
506 mockcollection.side_effect = lambda *args, **kwargs: CollectionMock(mock.MagicMock(), *args, **kwargs)
508 arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, loadingContext)
509 arvtool.formatgraph = None
510 it = arvtool.job({}, mock.MagicMock(), runtimeContext)
512 next(it).run(runtimeContext)
513 next(it).run(runtimeContext)
515 with open("tests/wf/echo-subwf.cwl") as f:
516 subwf = StripYAMLComments(f.read())
518 runner.api.jobs().create.assert_called_with(
519 body=JsonDiffMatcher({
520 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
521 'repository': 'arvados',
522 'script_version': 'master',
523 'script': 'crunchrunner',
524 'script_parameters': {
525 'tasks': [{'task.env': {
526 'HOME': '$(task.outdir)',
527 'TMPDIR': '$(task.tmpdir)'},
529 'workflow.cwl': '$(task.keep)/99999999999999999999999999999996+99/workflow.cwl',
530 'cwl.input.yml': '$(task.keep)/99999999999999999999999999999996+99/cwl.input.yml'
532 'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'],
533 'task.stdout': 'cwl.output.json'}]},
534 'runtime_constraints': {
535 'min_scratch_mb_per_node': 4096,
536 'min_cores_per_node': 3,
537 'docker_image': 'arvados/jobs',
538 'min_ram_mb_per_node': 1024
540 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
541 filters=[['repository', '=', 'arvados'],
542 ['script', '=', 'crunchrunner'],
543 ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
544 ['docker_image_locator', 'in docker', 'arvados/jobs']],
547 def test_default_work_api(self):
548 arvados_cwl.add_arv_hints()
550 api = mock.MagicMock()
551 api._rootDesc = copy.deepcopy(get_rootDesc())
552 del api._rootDesc.get('resources')['jobs']['methods']['create']
553 runner = arvados_cwl.executor.ArvCwlExecutor(api)
554 self.assertEqual(runner.work_api, 'containers')