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
22 from .test_container import CollectionMock
24 if not os.getenv('ARVADOS_DEBUG'):
25 logging.getLogger('arvados.cwl-runner').setLevel(logging.WARN)
26 logging.getLogger('arvados.arv-run').setLevel(logging.WARN)
28 class TestJob(unittest.TestCase):
30 # The test passes no builder.resources
31 # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
32 @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
33 def test_run(self, list_images_in_arv):
34 for enable_reuse in (True, False):
35 runner = mock.MagicMock()
36 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
37 runner.ignore_docker_for_reuse = False
38 runner.num_retries = 0
39 document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
41 list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
42 runner.api.collections().get().execute.return_value = {"portable_data_hash": "99999999999999999999999999999993+99"}
43 # Simulate reused job from another project so that we can check is a can_read
45 runner.api.jobs().create().execute.return_value = {
46 'state': 'Complete' if enable_reuse else 'Queued',
47 'owner_uuid': 'zzzzz-tpzed-yyyyyyyyyyyyyyy' if enable_reuse else 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
48 'uuid': 'zzzzz-819sb-yyyyyyyyyyyyyyy',
56 "arguments": [{"valueFrom": "$(runtime.outdir)"}],
58 "class": "CommandLineTool"
60 make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
61 collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
62 arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
63 basedir="", make_fs_access=make_fs_access, loader=Loader({}),
64 metadata={"cwlVersion": "v1.0"})
65 arvtool.formatgraph = None
66 for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
67 j.run(enable_reuse=enable_reuse)
68 runner.api.jobs().create.assert_called_with(
69 body=JsonDiffMatcher({
70 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
71 'runtime_constraints': {},
72 'script_parameters': {
74 'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
75 'command': ['ls', '$(task.outdir)']
78 'script_version': 'master',
79 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
80 'repository': 'arvados',
81 'script': 'crunchrunner',
82 'runtime_constraints': {
83 'docker_image': 'arvados/jobs',
84 'min_cores_per_node': 1,
85 'min_ram_mb_per_node': 1024,
86 'min_scratch_mb_per_node': 2048 # tmpdirSize + outdirSize
89 find_or_create=enable_reuse,
90 filters=[['repository', '=', 'arvados'],
91 ['script', '=', 'crunchrunner'],
92 ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
93 ['docker_image_locator', 'in docker', 'arvados/jobs']]
96 runner.api.links().create.assert_called_with(
97 body=JsonDiffMatcher({
98 'link_class': 'permission',
100 "tail_uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz",
101 "head_uuid": "zzzzz-819sb-yyyyyyyyyyyyyyy",
104 # Simulate an API excepction when trying to create a
105 # sharing link on the job
106 runner.api.links().create.side_effect = ApiError(
107 mock.MagicMock(return_value={'status': 403}),
109 j.run(enable_reuse=enable_reuse)
111 assert not runner.api.links().create.called
113 # The test passes some fields in builder.resources
114 # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
115 @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
116 def test_resource_requirements(self, list_images_in_arv):
117 runner = mock.MagicMock()
118 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
119 runner.ignore_docker_for_reuse = False
120 runner.num_retries = 0
121 arvados_cwl.add_arv_hints()
123 list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
124 runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
126 document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
133 "class": "ResourceRequirement",
138 "class": "http://arvados.org/cwl#RuntimeConstraints",
140 "outputDirType": "keep_output_dir"
142 "class": "http://arvados.org/cwl#APIRequirement",
145 "class": "http://arvados.org/cwl#ReuseRequirement",
150 "class": "CommandLineTool"
152 make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
153 collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
154 arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
155 make_fs_access=make_fs_access, loader=Loader({}),
156 metadata={"cwlVersion": "v1.0"})
157 arvtool.formatgraph = None
158 for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
159 j.run(enable_reuse=True)
160 runner.api.jobs().create.assert_called_with(
161 body=JsonDiffMatcher({
162 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
163 'runtime_constraints': {},
164 'script_parameters': {
166 'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
167 'task.keepTmpOutput': True,
171 'script_version': 'master',
172 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
173 'repository': 'arvados',
174 'script': 'crunchrunner',
175 'runtime_constraints': {
176 'docker_image': 'arvados/jobs',
177 'min_cores_per_node': 3,
178 'min_ram_mb_per_node': 3512, # ramMin + keep_cache
179 'min_scratch_mb_per_node': 5024, # tmpdirSize + outdirSize
180 'keep_cache_mb_per_task': 512
183 find_or_create=False,
184 filters=[['repository', '=', 'arvados'],
185 ['script', '=', 'crunchrunner'],
186 ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
187 ['docker_image_locator', 'in docker', 'arvados/jobs']])
189 @mock.patch("arvados.collection.CollectionReader")
190 def test_done(self, reader):
191 api = mock.MagicMock()
193 runner = mock.MagicMock()
195 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
196 runner.num_retries = 0
197 runner.ignore_docker_for_reuse = False
199 reader().open.return_value = StringIO.StringIO(
200 """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
201 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
202 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
204 api.collections().list().execute.side_effect = ({"items": []},
205 {"items": [{"manifest_text": "XYZ"}]},
207 {"items": [{"manifest_text": "ABC"}]})
209 arvjob = arvados_cwl.ArvadosJob(runner)
210 arvjob.name = "testjob"
211 arvjob.builder = mock.MagicMock()
212 arvjob.output_callback = mock.MagicMock()
213 arvjob.collect_outputs = mock.MagicMock()
214 arvjob.collect_outputs.return_value = {"out": "stuff"}
218 "output": "99999999999999999999999999999993+99",
219 "log": "99999999999999999999999999999994+99",
220 "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
223 api.collections().list.assert_has_calls([
225 # Output collection check
226 mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
227 ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
228 ['name', '=', 'Output 9999999 of testjob']]),
229 mock.call().execute(num_retries=0),
230 mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999993+99']],
231 select=['manifest_text']),
232 mock.call().execute(num_retries=0),
233 # Log collection's turn
234 mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
235 ['portable_data_hash', '=', '99999999999999999999999999999994+99'],
236 ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]),
237 mock.call().execute(num_retries=0),
238 mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999994+99']],
239 select=['manifest_text']),
240 mock.call().execute(num_retries=0)])
242 api.collections().create.assert_has_calls([
243 mock.call(ensure_unique_name=True,
244 body={'portable_data_hash': '99999999999999999999999999999993+99',
245 'manifest_text': 'XYZ',
246 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
247 'name': 'Output 9999999 of testjob'}),
248 mock.call().execute(num_retries=0),
249 mock.call(ensure_unique_name=True,
250 body={'portable_data_hash': '99999999999999999999999999999994+99',
251 'manifest_text': 'ABC',
252 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
253 'name': 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
254 mock.call().execute(num_retries=0),
257 arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
259 @mock.patch("arvados.collection.CollectionReader")
260 def test_done_use_existing_collection(self, reader):
261 api = mock.MagicMock()
263 runner = mock.MagicMock()
265 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
266 runner.num_retries = 0
268 reader().open.return_value = StringIO.StringIO(
269 """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
270 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
271 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
274 api.collections().list().execute.side_effect = (
275 {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},
276 {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},
279 arvjob = arvados_cwl.ArvadosJob(runner)
280 arvjob.name = "testjob"
281 arvjob.builder = mock.MagicMock()
282 arvjob.output_callback = mock.MagicMock()
283 arvjob.collect_outputs = mock.MagicMock()
284 arvjob.collect_outputs.return_value = {"out": "stuff"}
288 "output": "99999999999999999999999999999993+99",
289 "log": "99999999999999999999999999999994+99",
290 "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
293 api.collections().list.assert_has_calls([
296 mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
297 ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
298 ['name', '=', 'Output 9999999 of testjob']]),
299 mock.call().execute(num_retries=0),
301 mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
302 ['portable_data_hash', '=', '99999999999999999999999999999994+99'],
303 ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]),
304 mock.call().execute(num_retries=0)
307 self.assertFalse(api.collections().create.called)
309 arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
312 class TestWorkflow(unittest.TestCase):
313 # The test passes no builder.resources
314 # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
315 @mock.patch("arvados.collection.CollectionReader")
316 @mock.patch("arvados.collection.Collection")
317 @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
318 def test_run(self, list_images_in_arv, mockcollection, mockcollectionreader):
319 arvados_cwl.add_arv_hints()
321 api = mock.MagicMock()
322 api._rootDesc = get_rootDesc()
324 runner = arvados_cwl.ArvCwlRunner(api)
325 self.assertEqual(runner.work_api, 'jobs')
327 list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
328 runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
329 runner.api.collections().list().execute.return_vaulue = {"items": [{"portable_data_hash": "99999999999999999999999999999993+99"}]}
331 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
332 runner.ignore_docker_for_reuse = False
333 runner.num_retries = 0
334 document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
336 make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
337 collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
338 document_loader.fetcher_constructor = functools.partial(arvados_cwl.CollectionFetcher, api_client=api, fs_access=make_fs_access(""))
339 document_loader.fetcher = document_loader.fetcher_constructor(document_loader.cache, document_loader.session)
340 document_loader.fetch_text = document_loader.fetcher.fetch_text
341 document_loader.check_exists = document_loader.fetcher.check_exists
343 tool, metadata = document_loader.resolve_ref("tests/wf/scatter2.cwl")
344 metadata["cwlVersion"] = tool["cwlVersion"]
346 mockc = mock.MagicMock()
347 mockcollection.side_effect = lambda *args, **kwargs: CollectionMock(mockc, *args, **kwargs)
349 arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names,
350 basedir="", make_fs_access=make_fs_access, loader=document_loader,
351 makeTool=runner.arv_make_tool, metadata=metadata)
352 arvtool.formatgraph = None
353 it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access, tmp_outdir_prefix="")
357 with open("tests/wf/scatter2_subwf.cwl") as f:
358 subwf = StripYAMLComments(f.read())
360 runner.api.jobs().create.assert_called_with(
361 body=JsonDiffMatcher({
362 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
363 'repository': 'arvados',
364 'script_version': 'master',
365 'script': 'crunchrunner',
366 'script_parameters': {
367 'tasks': [{'task.env': {
368 'HOME': '$(task.outdir)',
369 'TMPDIR': '$(task.tmpdir)'},
371 'workflow.cwl': '$(task.keep)/99999999999999999999999999999996+99/workflow.cwl',
372 'cwl.input.yml': '$(task.keep)/99999999999999999999999999999996+99/cwl.input.yml'
374 'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'],
375 'task.stdout': 'cwl.output.json'}]},
376 'runtime_constraints': {
377 'min_scratch_mb_per_node': 2048,
378 'min_cores_per_node': 1,
379 'docker_image': 'arvados/jobs',
380 'min_ram_mb_per_node': 1024
382 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
383 filters=[['repository', '=', 'arvados'],
384 ['script', '=', 'crunchrunner'],
385 ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
386 ['docker_image_locator', 'in docker', 'arvados/jobs']],
389 mockc.open().__enter__().write.assert_has_calls([mock.call(subwf)])
390 mockc.open().__enter__().write.assert_has_calls([mock.call(
393 "basename": "token.txt",
395 "location": "/keep/99999999999999999999999999999999+118/token.txt"
400 # The test passes no builder.resources
401 # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
402 @mock.patch("arvados.collection.CollectionReader")
403 @mock.patch("arvados.collection.Collection")
404 @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
405 def test_overall_resource_singlecontainer(self, list_images_in_arv, mockcollection, mockcollectionreader):
406 arvados_cwl.add_arv_hints()
408 api = mock.MagicMock()
409 api._rootDesc = get_rootDesc()
411 runner = arvados_cwl.ArvCwlRunner(api)
412 self.assertEqual(runner.work_api, 'jobs')
414 list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
415 runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
416 runner.api.collections().list().execute.return_vaulue = {"items": [{"portable_data_hash": "99999999999999999999999999999993+99"}]}
418 runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
419 runner.ignore_docker_for_reuse = False
420 runner.num_retries = 0
421 document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
423 make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
424 collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
425 document_loader.fetcher_constructor = functools.partial(arvados_cwl.CollectionFetcher, api_client=api, fs_access=make_fs_access(""))
426 document_loader.fetcher = document_loader.fetcher_constructor(document_loader.cache, document_loader.session)
427 document_loader.fetch_text = document_loader.fetcher.fetch_text
428 document_loader.check_exists = document_loader.fetcher.check_exists
430 tool, metadata = document_loader.resolve_ref("tests/wf/echo-wf.cwl")
431 metadata["cwlVersion"] = tool["cwlVersion"]
433 mockcollection.side_effect = lambda *args, **kwargs: CollectionMock(mock.MagicMock(), *args, **kwargs)
435 arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names,
436 basedir="", make_fs_access=make_fs_access, loader=document_loader,
437 makeTool=runner.arv_make_tool, metadata=metadata)
438 arvtool.formatgraph = None
439 it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access, tmp_outdir_prefix="")
443 with open("tests/wf/echo-subwf.cwl") as f:
444 subwf = StripYAMLComments(f.read())
446 runner.api.jobs().create.assert_called_with(
447 body=JsonDiffMatcher({
448 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
449 'repository': 'arvados',
450 'script_version': 'master',
451 'script': 'crunchrunner',
452 'script_parameters': {
453 'tasks': [{'task.env': {
454 'HOME': '$(task.outdir)',
455 'TMPDIR': '$(task.tmpdir)'},
457 'workflow.cwl': '$(task.keep)/99999999999999999999999999999996+99/workflow.cwl',
458 'cwl.input.yml': '$(task.keep)/99999999999999999999999999999996+99/cwl.input.yml'
460 'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'],
461 'task.stdout': 'cwl.output.json'}]},
462 'runtime_constraints': {
463 'min_scratch_mb_per_node': 4096,
464 'min_cores_per_node': 3,
465 'docker_image': 'arvados/jobs',
466 'min_ram_mb_per_node': 1024
468 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
469 filters=[['repository', '=', 'arvados'],
470 ['script', '=', 'crunchrunner'],
471 ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
472 ['docker_image_locator', 'in docker', 'arvados/jobs']],
475 def test_default_work_api(self):
476 arvados_cwl.add_arv_hints()
478 api = mock.MagicMock()
479 api._rootDesc = copy.deepcopy(get_rootDesc())
480 del api._rootDesc.get('resources')['jobs']['methods']['create']
481 runner = arvados_cwl.ArvCwlRunner(api)
482 self.assertEqual(runner.work_api, 'containers')