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": "$(runtime.outdir)"}],
57 "class": "CommandLineTool"
59 make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
60 collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
61 arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
62 basedir="", make_fs_access=make_fs_access, loader=Loader({}),
63 metadata={"cwlVersion": "v1.0"})
64 arvtool.formatgraph = None
65 for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
66 j.run(enable_reuse=enable_reuse)
67 runner.api.jobs().create.assert_called_with(
68 body=JsonDiffMatcher({
69 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
70 'runtime_constraints': {},
71 'script_parameters': {
73 'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
74 'command': ['ls', '$(task.outdir)']
77 'script_version': 'master',
78 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
79 'repository': 'arvados',
80 'script': 'crunchrunner',
81 'runtime_constraints': {
82 'docker_image': 'arvados/jobs',
83 'min_cores_per_node': 1,
84 'min_ram_mb_per_node': 1024,
85 'min_scratch_mb_per_node': 2048 # tmpdirSize + outdirSize
88 find_or_create=enable_reuse,
89 filters=[['repository', '=', 'arvados'],
90 ['script', '=', 'crunchrunner'],
91 ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
92 ['docker_image_locator', 'in docker', 'arvados/jobs']]
95 runner.api.links().create.assert_called_with(
96 body=JsonDiffMatcher({
97 'link_class': 'permission',
99 "tail_uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz",
100 "head_uuid": "zzzzz-819sb-yyyyyyyyyyyyyyy",
103 # Simulate an API excepction when trying to create a
104 # sharing link on the job
105 runner.api.links().create.side_effect = ApiError(
106 mock.MagicMock(return_value={'status': 403}),
108 j.run(enable_reuse=enable_reuse)
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 metadata={"cwlVersion": "v1.0"})
156 arvtool.formatgraph = None
157 for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
158 j.run(enable_reuse=True)
159 runner.api.jobs().create.assert_called_with(
160 body=JsonDiffMatcher({
161 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
162 'runtime_constraints': {},
163 'script_parameters': {
165 'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
166 'task.keepTmpOutput': True,
170 'script_version': 'master',
171 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
172 'repository': 'arvados',
173 'script': 'crunchrunner',
174 'runtime_constraints': {
175 'docker_image': 'arvados/jobs',
176 'min_cores_per_node': 3,
177 'min_ram_mb_per_node': 3512, # ramMin + keep_cache
178 'min_scratch_mb_per_node': 5024, # tmpdirSize + outdirSize
179 'keep_cache_mb_per_task': 512
182 find_or_create=False,
183 filters=[['repository', '=', 'arvados'],
184 ['script', '=', 'crunchrunner'],
185 ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
186 ['docker_image_locator', 'in docker', 'arvados/jobs']])
188 @mock.patch("arvados.collection.CollectionReader")
189 def test_done(self, reader):
190 api = mock.MagicMock()
192 runner = mock.MagicMock()
194 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
195 runner.num_retries = 0
196 runner.ignore_docker_for_reuse = False
198 reader().open.return_value = StringIO.StringIO(
199 """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
200 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
201 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
203 api.collections().list().execute.side_effect = ({"items": []},
204 {"items": [{"manifest_text": "XYZ"}]},
206 {"items": [{"manifest_text": "ABC"}]})
208 arvjob = arvados_cwl.ArvadosJob(runner)
209 arvjob.name = "testjob"
210 arvjob.builder = mock.MagicMock()
211 arvjob.output_callback = mock.MagicMock()
212 arvjob.collect_outputs = mock.MagicMock()
213 arvjob.collect_outputs.return_value = {"out": "stuff"}
217 "output": "99999999999999999999999999999993+99",
218 "log": "99999999999999999999999999999994+99",
219 "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
222 api.collections().list.assert_has_calls([
224 # Output collection check
225 mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
226 ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
227 ['name', '=', 'Output 9999999 of testjob']]),
228 mock.call().execute(num_retries=0),
229 mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999993+99']],
230 select=['manifest_text']),
231 mock.call().execute(num_retries=0),
232 # Log collection's turn
233 mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
234 ['portable_data_hash', '=', '99999999999999999999999999999994+99'],
235 ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]),
236 mock.call().execute(num_retries=0),
237 mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999994+99']],
238 select=['manifest_text']),
239 mock.call().execute(num_retries=0)])
241 api.collections().create.assert_has_calls([
242 mock.call(ensure_unique_name=True,
243 body={'portable_data_hash': '99999999999999999999999999999993+99',
244 'manifest_text': 'XYZ',
245 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
246 'name': 'Output 9999999 of testjob'}),
247 mock.call().execute(num_retries=0),
248 mock.call(ensure_unique_name=True,
249 body={'portable_data_hash': '99999999999999999999999999999994+99',
250 'manifest_text': 'ABC',
251 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
252 'name': 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
253 mock.call().execute(num_retries=0),
256 arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
258 @mock.patch("arvados.collection.CollectionReader")
259 def test_done_use_existing_collection(self, reader):
260 api = mock.MagicMock()
262 runner = mock.MagicMock()
264 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
265 runner.num_retries = 0
267 reader().open.return_value = StringIO.StringIO(
268 """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
269 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
270 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
273 api.collections().list().execute.side_effect = (
274 {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},
275 {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},
278 arvjob = arvados_cwl.ArvadosJob(runner)
279 arvjob.name = "testjob"
280 arvjob.builder = mock.MagicMock()
281 arvjob.output_callback = mock.MagicMock()
282 arvjob.collect_outputs = mock.MagicMock()
283 arvjob.collect_outputs.return_value = {"out": "stuff"}
287 "output": "99999999999999999999999999999993+99",
288 "log": "99999999999999999999999999999994+99",
289 "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
292 api.collections().list.assert_has_calls([
295 mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
296 ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
297 ['name', '=', 'Output 9999999 of testjob']]),
298 mock.call().execute(num_retries=0),
300 mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
301 ['portable_data_hash', '=', '99999999999999999999999999999994+99'],
302 ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]),
303 mock.call().execute(num_retries=0)
306 self.assertFalse(api.collections().create.called)
308 arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
311 class TestWorkflow(unittest.TestCase):
312 # The test passes no builder.resources
313 # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
314 @mock.patch("arvados.collection.CollectionReader")
315 @mock.patch("arvados.collection.Collection")
316 @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
317 def test_run(self, list_images_in_arv, mockcollection, mockcollectionreader):
318 arvados_cwl.add_arv_hints()
320 api = mock.MagicMock()
321 api._rootDesc = get_rootDesc()
323 runner = arvados_cwl.ArvCwlRunner(api)
324 self.assertEqual(runner.work_api, 'jobs')
326 list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
327 runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
328 runner.api.collections().list().execute.return_vaulue = {"items": [{"portable_data_hash": "99999999999999999999999999999993+99"}]}
330 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
331 runner.ignore_docker_for_reuse = False
332 runner.num_retries = 0
333 document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
335 make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
336 collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
337 document_loader.fetcher_constructor = functools.partial(arvados_cwl.CollectionFetcher, api_client=api, fs_access=make_fs_access(""))
338 document_loader.fetcher = document_loader.fetcher_constructor(document_loader.cache, document_loader.session)
339 document_loader.fetch_text = document_loader.fetcher.fetch_text
340 document_loader.check_exists = document_loader.fetcher.check_exists
342 tool, metadata = document_loader.resolve_ref("tests/wf/scatter2.cwl")
343 metadata["cwlVersion"] = tool["cwlVersion"]
345 mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118"
347 arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names,
348 basedir="", make_fs_access=make_fs_access, loader=document_loader,
349 makeTool=runner.arv_make_tool, metadata=metadata)
350 arvtool.formatgraph = None
351 it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access, tmp_outdir_prefix="")
355 with open("tests/wf/scatter2_subwf.cwl") as f:
356 subwf = StripYAMLComments(f.read())
358 runner.api.jobs().create.assert_called_with(
359 body=JsonDiffMatcher({
360 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
361 'repository': 'arvados',
362 'script_version': 'master',
363 'script': 'crunchrunner',
364 'script_parameters': {
365 'tasks': [{'task.env': {
366 'HOME': '$(task.outdir)',
367 'TMPDIR': '$(task.tmpdir)'},
369 'workflow.cwl': '$(task.keep)/99999999999999999999999999999999+118/workflow.cwl',
370 'cwl.input.yml': '$(task.keep)/99999999999999999999999999999999+118/cwl.input.yml'
372 'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'],
373 'task.stdout': 'cwl.output.json'}]},
374 'runtime_constraints': {
375 'min_scratch_mb_per_node': 2048,
376 'min_cores_per_node': 1,
377 'docker_image': 'arvados/jobs',
378 'min_ram_mb_per_node': 1024
380 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
381 filters=[['repository', '=', 'arvados'],
382 ['script', '=', 'crunchrunner'],
383 ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
384 ['docker_image_locator', 'in docker', 'arvados/jobs']],
387 mockcollection().open().__enter__().write.assert_has_calls([mock.call(subwf)])
388 mockcollection().open().__enter__().write.assert_has_calls([mock.call(
391 "basename": "token.txt",
393 "location": "/keep/99999999999999999999999999999999+118/token.txt"
398 # The test passes no builder.resources
399 # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
400 @mock.patch("arvados.collection.CollectionReader")
401 @mock.patch("arvados.collection.Collection")
402 @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
403 def test_overall_resource_singlecontainer(self, list_images_in_arv, mockcollection, mockcollectionreader):
404 arvados_cwl.add_arv_hints()
406 api = mock.MagicMock()
407 api._rootDesc = get_rootDesc()
409 runner = arvados_cwl.ArvCwlRunner(api)
410 self.assertEqual(runner.work_api, 'jobs')
412 list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
413 runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
414 runner.api.collections().list().execute.return_vaulue = {"items": [{"portable_data_hash": "99999999999999999999999999999993+99"}]}
416 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
417 runner.ignore_docker_for_reuse = False
418 runner.num_retries = 0
419 document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
421 make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
422 collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
423 document_loader.fetcher_constructor = functools.partial(arvados_cwl.CollectionFetcher, api_client=api, fs_access=make_fs_access(""))
424 document_loader.fetcher = document_loader.fetcher_constructor(document_loader.cache, document_loader.session)
425 document_loader.fetch_text = document_loader.fetcher.fetch_text
426 document_loader.check_exists = document_loader.fetcher.check_exists
428 tool, metadata = document_loader.resolve_ref("tests/wf/echo-wf.cwl")
429 metadata["cwlVersion"] = tool["cwlVersion"]
431 mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118"
433 arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names,
434 basedir="", make_fs_access=make_fs_access, loader=document_loader,
435 makeTool=runner.arv_make_tool, metadata=metadata)
436 arvtool.formatgraph = None
437 it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access, tmp_outdir_prefix="")
441 with open("tests/wf/echo-subwf.cwl") as f:
442 subwf = StripYAMLComments(f.read())
444 runner.api.jobs().create.assert_called_with(
445 body=JsonDiffMatcher({
446 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
447 'repository': 'arvados',
448 'script_version': 'master',
449 'script': 'crunchrunner',
450 'script_parameters': {
451 'tasks': [{'task.env': {
452 'HOME': '$(task.outdir)',
453 'TMPDIR': '$(task.tmpdir)'},
455 'workflow.cwl': '$(task.keep)/99999999999999999999999999999999+118/workflow.cwl',
456 'cwl.input.yml': '$(task.keep)/99999999999999999999999999999999+118/cwl.input.yml'
458 'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'],
459 'task.stdout': 'cwl.output.json'}]},
460 'runtime_constraints': {
461 'min_scratch_mb_per_node': 4096,
462 'min_cores_per_node': 3,
463 'docker_image': 'arvados/jobs',
464 'min_ram_mb_per_node': 1024
466 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
467 filters=[['repository', '=', 'arvados'],
468 ['script', '=', 'crunchrunner'],
469 ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
470 ['docker_image_locator', 'in docker', 'arvados/jobs']],
473 def test_default_work_api(self):
474 arvados_cwl.add_arv_hints()
476 api = mock.MagicMock()
477 api._rootDesc = copy.deepcopy(get_rootDesc())
478 del api._rootDesc.get('resources')['jobs']['methods']['create']
479 runner = arvados_cwl.ArvCwlRunner(api)
480 self.assertEqual(runner.work_api, 'containers')