1 # Copyright (C) The Arvados Authors. All rights reserved.
3 # SPDX-License-Identifier: Apache-2.0
16 import cwltool.process
17 from arvados.errors import ApiError
18 from schema_salad.ref_resolver import Loader
19 from schema_salad.sourceline import cmap
20 from .mock_discovery import get_rootDesc
21 from .matcher import JsonDiffMatcher, StripYAMLComments
23 if not os.getenv('ARVADOS_DEBUG'):
24 logging.getLogger('arvados.cwl-runner').setLevel(logging.WARN)
25 logging.getLogger('arvados.arv-run').setLevel(logging.WARN)
27 class TestJob(unittest.TestCase):
29 # The test passes no builder.resources
30 # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
31 @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
32 def test_run(self, list_images_in_arv):
33 for enable_reuse in (True, False):
34 runner = mock.MagicMock()
35 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
36 runner.ignore_docker_for_reuse = False
37 runner.num_retries = 0
38 document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
40 list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
41 runner.api.collections().get().execute.return_value = {"portable_data_hash": "99999999999999999999999999999993+99"}
42 # Simulate reused job from another project so that we can check is a can_read
44 runner.api.jobs().create().execute.return_value = {
45 'state': 'Complete' if enable_reuse else 'Queued',
46 'owner_uuid': 'zzzzz-tpzed-yyyyyyyyyyyyyyy' if enable_reuse else 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
47 'uuid': 'zzzzz-819sb-yyyyyyyyyyyyyyy',
55 "arguments": [{"valueFrom": "$(runtime.outdir)"}],
57 "class": "CommandLineTool"
59 make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
60 collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
61 arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
62 basedir="", make_fs_access=make_fs_access, loader=Loader({}))
63 arvtool.formatgraph = None
64 for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
65 j.run(enable_reuse=enable_reuse)
66 runner.api.jobs().create.assert_called_with(
67 body=JsonDiffMatcher({
68 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
69 'runtime_constraints': {},
70 'script_parameters': {
72 'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
73 'command': ['ls', '$(task.outdir)']
76 'script_version': 'master',
77 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
78 'repository': 'arvados',
79 'script': 'crunchrunner',
80 'runtime_constraints': {
81 'docker_image': 'arvados/jobs',
82 'min_cores_per_node': 1,
83 'min_ram_mb_per_node': 1024,
84 'min_scratch_mb_per_node': 2048 # tmpdirSize + outdirSize
87 find_or_create=enable_reuse,
88 filters=[['repository', '=', 'arvados'],
89 ['script', '=', 'crunchrunner'],
90 ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
91 ['docker_image_locator', 'in docker', 'arvados/jobs']]
94 runner.api.links().create.assert_called_with(
95 body=JsonDiffMatcher({
96 'link_class': 'permission',
98 "tail_uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz",
99 "head_uuid": "zzzzz-819sb-yyyyyyyyyyyyyyy",
102 # Simulate an API excepction when trying to create a
103 # sharing link on the job
104 runner.api.links().create.side_effect = ApiError(
105 mock.MagicMock(return_value={'status': 403}),
107 j.run(enable_reuse=enable_reuse)
108 j.output_callback.assert_called_with({}, 'success')
110 assert not runner.api.links().create.called
112 # The test passes some fields in builder.resources
113 # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
114 @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
115 def test_resource_requirements(self, list_images_in_arv):
116 runner = mock.MagicMock()
117 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
118 runner.ignore_docker_for_reuse = False
119 runner.num_retries = 0
120 arvados_cwl.add_arv_hints()
122 list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
123 runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
125 document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
132 "class": "ResourceRequirement",
137 "class": "http://arvados.org/cwl#RuntimeConstraints",
139 "outputDirType": "keep_output_dir"
141 "class": "http://arvados.org/cwl#APIRequirement",
144 "class": "http://arvados.org/cwl#ReuseRequirement",
149 "class": "CommandLineTool"
151 make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
152 collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
153 arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
154 make_fs_access=make_fs_access, loader=Loader({}))
155 arvtool.formatgraph = None
156 for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
157 j.run(enable_reuse=True)
158 runner.api.jobs().create.assert_called_with(
159 body=JsonDiffMatcher({
160 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
161 'runtime_constraints': {},
162 'script_parameters': {
164 'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
165 'task.keepTmpOutput': True,
169 'script_version': 'master',
170 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
171 'repository': 'arvados',
172 'script': 'crunchrunner',
173 'runtime_constraints': {
174 'docker_image': 'arvados/jobs',
175 'min_cores_per_node': 3,
176 'min_ram_mb_per_node': 3512, # ramMin + keep_cache
177 'min_scratch_mb_per_node': 5024, # tmpdirSize + outdirSize
178 'keep_cache_mb_per_task': 512
181 find_or_create=False,
182 filters=[['repository', '=', 'arvados'],
183 ['script', '=', 'crunchrunner'],
184 ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
185 ['docker_image_locator', 'in docker', 'arvados/jobs']])
187 @mock.patch("arvados.collection.CollectionReader")
188 def test_done(self, reader):
189 api = mock.MagicMock()
191 runner = mock.MagicMock()
193 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
194 runner.num_retries = 0
195 runner.ignore_docker_for_reuse = False
197 reader().open.return_value = StringIO.StringIO(
198 """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
199 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
200 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
202 api.collections().list().execute.side_effect = ({"items": []},
203 {"items": [{"manifest_text": "XYZ"}]},
205 {"items": [{"manifest_text": "ABC"}]})
207 arvjob = arvados_cwl.ArvadosJob(runner)
208 arvjob.name = "testjob"
209 arvjob.builder = mock.MagicMock()
210 arvjob.output_callback = mock.MagicMock()
211 arvjob.collect_outputs = mock.MagicMock()
212 arvjob.collect_outputs.return_value = {"out": "stuff"}
216 "output": "99999999999999999999999999999993+99",
217 "log": "99999999999999999999999999999994+99",
218 "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
221 api.collections().list.assert_has_calls([
223 # Output collection check
224 mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
225 ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
226 ['name', '=', 'Output 9999999 of testjob']]),
227 mock.call().execute(num_retries=0),
228 mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999993+99']],
229 select=['manifest_text']),
230 mock.call().execute(num_retries=0),
231 # Log collection's turn
232 mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
233 ['portable_data_hash', '=', '99999999999999999999999999999994+99'],
234 ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]),
235 mock.call().execute(num_retries=0),
236 mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999994+99']],
237 select=['manifest_text']),
238 mock.call().execute(num_retries=0)])
240 api.collections().create.assert_has_calls([
241 mock.call(ensure_unique_name=True,
242 body={'portable_data_hash': '99999999999999999999999999999993+99',
243 'manifest_text': 'XYZ',
244 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
245 'name': 'Output 9999999 of testjob'}),
246 mock.call().execute(num_retries=0),
247 mock.call(ensure_unique_name=True,
248 body={'portable_data_hash': '99999999999999999999999999999994+99',
249 'manifest_text': 'ABC',
250 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
251 'name': 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
252 mock.call().execute(num_retries=0),
255 arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
257 @mock.patch("arvados.collection.CollectionReader")
258 def test_done_use_existing_collection(self, reader):
259 api = mock.MagicMock()
261 runner = mock.MagicMock()
263 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
264 runner.num_retries = 0
266 reader().open.return_value = StringIO.StringIO(
267 """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
268 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
269 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
272 api.collections().list().execute.side_effect = (
273 {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},
274 {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},
277 arvjob = arvados_cwl.ArvadosJob(runner)
278 arvjob.name = "testjob"
279 arvjob.builder = mock.MagicMock()
280 arvjob.output_callback = mock.MagicMock()
281 arvjob.collect_outputs = mock.MagicMock()
282 arvjob.collect_outputs.return_value = {"out": "stuff"}
286 "output": "99999999999999999999999999999993+99",
287 "log": "99999999999999999999999999999994+99",
288 "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
291 api.collections().list.assert_has_calls([
294 mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
295 ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
296 ['name', '=', 'Output 9999999 of testjob']]),
297 mock.call().execute(num_retries=0),
299 mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
300 ['portable_data_hash', '=', '99999999999999999999999999999994+99'],
301 ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]),
302 mock.call().execute(num_retries=0)
305 self.assertFalse(api.collections().create.called)
307 arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
310 class TestWorkflow(unittest.TestCase):
311 # The test passes no builder.resources
312 # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
313 @mock.patch("arvados.collection.CollectionReader")
314 @mock.patch("arvados.collection.Collection")
315 @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
316 def test_run(self, list_images_in_arv, mockcollection, mockcollectionreader):
317 arvados_cwl.add_arv_hints()
319 api = mock.MagicMock()
320 api._rootDesc = get_rootDesc()
322 runner = arvados_cwl.ArvCwlRunner(api)
323 self.assertEqual(runner.work_api, 'jobs')
325 list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
326 runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
327 runner.api.collections().list().execute.return_vaulue = {"items": [{"portable_data_hash": "99999999999999999999999999999993+99"}]}
329 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
330 runner.ignore_docker_for_reuse = False
331 runner.num_retries = 0
332 document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
334 make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
335 collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
336 document_loader.fetcher_constructor = functools.partial(arvados_cwl.CollectionFetcher, api_client=api, fs_access=make_fs_access(""))
337 document_loader.fetcher = document_loader.fetcher_constructor(document_loader.cache, document_loader.session)
338 document_loader.fetch_text = document_loader.fetcher.fetch_text
339 document_loader.check_exists = document_loader.fetcher.check_exists
341 tool, metadata = document_loader.resolve_ref("tests/wf/scatter2.cwl")
342 metadata["cwlVersion"] = tool["cwlVersion"]
344 mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118"
346 arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names,
347 basedir="", make_fs_access=make_fs_access, loader=document_loader,
348 makeTool=runner.arv_make_tool, metadata=metadata)
349 arvtool.formatgraph = None
350 it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access)
354 with open("tests/wf/scatter2_subwf.cwl") as f:
355 subwf = StripYAMLComments(f.read())
357 runner.api.jobs().create.assert_called_with(
358 body=JsonDiffMatcher({
359 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
360 'repository': 'arvados',
361 'script_version': 'master',
362 'script': 'crunchrunner',
363 'script_parameters': {
364 'tasks': [{'task.env': {
365 'HOME': '$(task.outdir)',
366 'TMPDIR': '$(task.tmpdir)'},
368 'workflow.cwl': '$(task.keep)/99999999999999999999999999999999+118/workflow.cwl',
369 'cwl.input.yml': '$(task.keep)/99999999999999999999999999999999+118/cwl.input.yml'
371 'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'],
372 'task.stdout': 'cwl.output.json'}]},
373 'runtime_constraints': {
374 'min_scratch_mb_per_node': 2048,
375 'min_cores_per_node': 1,
376 'docker_image': 'arvados/jobs',
377 'min_ram_mb_per_node': 1024
379 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
380 filters=[['repository', '=', 'arvados'],
381 ['script', '=', 'crunchrunner'],
382 ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
383 ['docker_image_locator', 'in docker', 'arvados/jobs']],
386 mockcollection().open().__enter__().write.assert_has_calls([mock.call(subwf)])
387 mockcollection().open().__enter__().write.assert_has_calls([mock.call(
390 "basename": "token.txt",
392 "location": "/keep/99999999999999999999999999999999+118/token.txt"
397 # The test passes no builder.resources
398 # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
399 @mock.patch("arvados.collection.CollectionReader")
400 @mock.patch("arvados.collection.Collection")
401 @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
402 def test_overall_resource_singlecontainer(self, list_images_in_arv, mockcollection, mockcollectionreader):
403 arvados_cwl.add_arv_hints()
405 api = mock.MagicMock()
406 api._rootDesc = get_rootDesc()
408 runner = arvados_cwl.ArvCwlRunner(api)
409 self.assertEqual(runner.work_api, 'jobs')
411 list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
412 runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
413 runner.api.collections().list().execute.return_vaulue = {"items": [{"portable_data_hash": "99999999999999999999999999999993+99"}]}
415 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
416 runner.ignore_docker_for_reuse = False
417 runner.num_retries = 0
418 document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
420 make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
421 collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
422 document_loader.fetcher_constructor = functools.partial(arvados_cwl.CollectionFetcher, api_client=api, fs_access=make_fs_access(""))
423 document_loader.fetcher = document_loader.fetcher_constructor(document_loader.cache, document_loader.session)
424 document_loader.fetch_text = document_loader.fetcher.fetch_text
425 document_loader.check_exists = document_loader.fetcher.check_exists
427 tool, metadata = document_loader.resolve_ref("tests/wf/echo-wf.cwl")
428 metadata["cwlVersion"] = tool["cwlVersion"]
430 mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118"
432 arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names,
433 basedir="", make_fs_access=make_fs_access, loader=document_loader,
434 makeTool=runner.arv_make_tool, metadata=metadata)
435 arvtool.formatgraph = None
436 it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access)
440 with open("tests/wf/echo-subwf.cwl") as f:
441 subwf = StripYAMLComments(f.read())
443 runner.api.jobs().create.assert_called_with(
444 body=JsonDiffMatcher({
445 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
446 'repository': 'arvados',
447 'script_version': 'master',
448 'script': 'crunchrunner',
449 'script_parameters': {
450 'tasks': [{'task.env': {
451 'HOME': '$(task.outdir)',
452 'TMPDIR': '$(task.tmpdir)'},
454 'workflow.cwl': '$(task.keep)/99999999999999999999999999999999+118/workflow.cwl',
455 'cwl.input.yml': '$(task.keep)/99999999999999999999999999999999+118/cwl.input.yml'
457 'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'],
458 'task.stdout': 'cwl.output.json'}]},
459 'runtime_constraints': {
460 'min_scratch_mb_per_node': 4096,
461 'min_cores_per_node': 3,
462 'docker_image': 'arvados/jobs',
463 'min_ram_mb_per_node': 1024
465 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
466 filters=[['repository', '=', 'arvados'],
467 ['script', '=', 'crunchrunner'],
468 ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
469 ['docker_image_locator', 'in docker', 'arvados/jobs']],
472 def test_default_work_api(self):
473 arvados_cwl.add_arv_hints()
475 api = mock.MagicMock()
476 api._rootDesc = copy.deepcopy(get_rootDesc())
477 del api._rootDesc.get('resources')['jobs']['methods']['create']
478 runner = arvados_cwl.ArvCwlRunner(api)
479 self.assertEqual(runner.work_api, 'containers')