2411: Add copyright notices to everything.
[arvados.git] / sdk / cwl / tests / test_job.py
1 # Copyright (C) The Arvados Authors. All rights reserved.
2 #
3 # SPDX-License-Identifier: Apache-2.0
4
5 import functools
6 import json
7 import logging
8 import mock
9 import os
10 import unittest
11 import copy
12 import StringIO
13
14 import arvados
15 import arvados_cwl
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
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)
26
27 class TestJob(unittest.TestCase):
28
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")
39
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
43             # link is added.
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',
48                 'output': None,
49             }
50
51             tool = cmap({
52                 "inputs": [],
53                 "outputs": [],
54                 "baseCommand": "ls",
55                 "arguments": [{"valueFrom": "$(runtime.outdir)"}]
56             })
57             make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
58                                          collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
59             arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
60                                                      basedir="", make_fs_access=make_fs_access, loader=Loader({}))
61             arvtool.formatgraph = None
62             for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
63                 j.run(enable_reuse=enable_reuse)
64                 runner.api.jobs().create.assert_called_with(
65                     body=JsonDiffMatcher({
66                         'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
67                         'runtime_constraints': {},
68                         'script_parameters': {
69                             'tasks': [{
70                                 'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
71                                 'command': ['ls', '$(task.outdir)']
72                             }],
73                         },
74                         'script_version': 'master',
75                         'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
76                         'repository': 'arvados',
77                         'script': 'crunchrunner',
78                         'runtime_constraints': {
79                             'docker_image': 'arvados/jobs',
80                             'min_cores_per_node': 1,
81                             'min_ram_mb_per_node': 1024,
82                             'min_scratch_mb_per_node': 2048 # tmpdirSize + outdirSize
83                         }
84                     }),
85                     find_or_create=enable_reuse,
86                     filters=[['repository', '=', 'arvados'],
87                              ['script', '=', 'crunchrunner'],
88                              ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
89                              ['docker_image_locator', 'in docker', 'arvados/jobs']]
90                 )
91                 if enable_reuse:
92                     runner.api.links().create.assert_called_with(
93                         body=JsonDiffMatcher({
94                             'link_class': 'permission',
95                             'name': 'can_read',
96                             "tail_uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz",
97                             "head_uuid": "zzzzz-819sb-yyyyyyyyyyyyyyy",
98                         })
99                     )
100                     # Simulate an API excepction when trying to create a
101                     # sharing link on the job
102                     runner.api.links().create.side_effect = ApiError(
103                         mock.MagicMock(return_value={'status': 403}),
104                         'Permission denied')
105                     j.run(enable_reuse=enable_reuse)
106                     j.output_callback.assert_called_with({}, 'success')
107                 else:
108                     assert not runner.api.links().create.called
109
110     # The test passes some fields in builder.resources
111     # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
112     @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
113     def test_resource_requirements(self, list_images_in_arv):
114         runner = mock.MagicMock()
115         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
116         runner.ignore_docker_for_reuse = False
117         runner.num_retries = 0
118         arvados_cwl.add_arv_hints()
119
120         list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
121         runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
122
123         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
124
125
126         tool = {
127             "inputs": [],
128             "outputs": [],
129             "hints": [{
130                 "class": "ResourceRequirement",
131                 "coresMin": 3,
132                 "ramMin": 3000,
133                 "tmpdirMin": 4000
134             }, {
135                 "class": "http://arvados.org/cwl#RuntimeConstraints",
136                 "keep_cache": 512,
137                 "outputDirType": "keep_output_dir"
138             }, {
139                 "class": "http://arvados.org/cwl#APIRequirement",
140             }],
141             "baseCommand": "ls"
142         }
143         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
144                                          collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
145         arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
146                                                  make_fs_access=make_fs_access, loader=Loader({}))
147         arvtool.formatgraph = None
148         for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
149             j.run()
150         runner.api.jobs().create.assert_called_with(
151             body=JsonDiffMatcher({
152                 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
153                 'runtime_constraints': {},
154                 'script_parameters': {
155                     'tasks': [{
156                         'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
157                         'task.keepTmpOutput': True,
158                         'command': ['ls']
159                     }]
160             },
161             'script_version': 'master',
162                 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
163                 'repository': 'arvados',
164                 'script': 'crunchrunner',
165                 'runtime_constraints': {
166                     'docker_image': 'arvados/jobs',
167                     'min_cores_per_node': 3,
168                     'min_ram_mb_per_node': 3512,     # ramMin + keep_cache
169                     'min_scratch_mb_per_node': 5024, # tmpdirSize + outdirSize
170                     'keep_cache_mb_per_task': 512
171                 }
172             }),
173             find_or_create=True,
174             filters=[['repository', '=', 'arvados'],
175                      ['script', '=', 'crunchrunner'],
176                      ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
177                      ['docker_image_locator', 'in docker', 'arvados/jobs']])
178
179     @mock.patch("arvados.collection.CollectionReader")
180     def test_done(self, reader):
181         api = mock.MagicMock()
182
183         runner = mock.MagicMock()
184         runner.api = api
185         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
186         runner.num_retries = 0
187         runner.ignore_docker_for_reuse = False
188
189         reader().open.return_value = StringIO.StringIO(
190             """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
191 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
192 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
193         """)
194         api.collections().list().execute.side_effect = ({"items": []},
195                                                         {"items": [{"manifest_text": "XYZ"}]},
196                                                         {"items": []},
197                                                         {"items": [{"manifest_text": "ABC"}]})
198
199         arvjob = arvados_cwl.ArvadosJob(runner)
200         arvjob.name = "testjob"
201         arvjob.builder = mock.MagicMock()
202         arvjob.output_callback = mock.MagicMock()
203         arvjob.collect_outputs = mock.MagicMock()
204         arvjob.collect_outputs.return_value = {"out": "stuff"}
205
206         arvjob.done({
207             "state": "Complete",
208             "output": "99999999999999999999999999999993+99",
209             "log": "99999999999999999999999999999994+99",
210             "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
211         })
212
213         api.collections().list.assert_has_calls([
214             mock.call(),
215             # Output collection check
216             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
217                           ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
218                           ['name', '=', 'Output 9999999 of testjob']]),
219             mock.call().execute(num_retries=0),
220             mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999993+99']],
221                  select=['manifest_text']),
222             mock.call().execute(num_retries=0),
223             # Log collection's turn
224             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
225                           ['portable_data_hash', '=', '99999999999999999999999999999994+99'],
226                           ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]),
227             mock.call().execute(num_retries=0),
228             mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999994+99']],
229                  select=['manifest_text']),
230             mock.call().execute(num_retries=0)])
231
232         api.collections().create.assert_has_calls([
233             mock.call(ensure_unique_name=True,
234                       body={'portable_data_hash': '99999999999999999999999999999993+99',
235                             'manifest_text': 'XYZ',
236                             'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
237                             'name': 'Output 9999999 of testjob'}),
238             mock.call().execute(num_retries=0),
239             mock.call(ensure_unique_name=True,
240                       body={'portable_data_hash': '99999999999999999999999999999994+99',
241                             'manifest_text': 'ABC',
242                             'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
243                             'name': 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
244             mock.call().execute(num_retries=0),
245         ])
246
247         arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
248
249     @mock.patch("arvados.collection.CollectionReader")
250     def test_done_use_existing_collection(self, reader):
251         api = mock.MagicMock()
252
253         runner = mock.MagicMock()
254         runner.api = api
255         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
256         runner.num_retries = 0
257
258         reader().open.return_value = StringIO.StringIO(
259             """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
260 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
261 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
262         """)
263
264         api.collections().list().execute.side_effect = (
265             {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},
266             {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},
267         )
268
269         arvjob = arvados_cwl.ArvadosJob(runner)
270         arvjob.name = "testjob"
271         arvjob.builder = mock.MagicMock()
272         arvjob.output_callback = mock.MagicMock()
273         arvjob.collect_outputs = mock.MagicMock()
274         arvjob.collect_outputs.return_value = {"out": "stuff"}
275
276         arvjob.done({
277             "state": "Complete",
278             "output": "99999999999999999999999999999993+99",
279             "log": "99999999999999999999999999999994+99",
280             "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
281         })
282
283         api.collections().list.assert_has_calls([
284             mock.call(),
285             # Output collection
286             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
287                                ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
288                                ['name', '=', 'Output 9999999 of testjob']]),
289             mock.call().execute(num_retries=0),
290             # Log collection
291             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
292                                ['portable_data_hash', '=', '99999999999999999999999999999994+99'],
293                                ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]),
294             mock.call().execute(num_retries=0)
295         ])
296
297         self.assertFalse(api.collections().create.called)
298
299         arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
300
301
302 class TestWorkflow(unittest.TestCase):
303     # The test passes no builder.resources
304     # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
305     @mock.patch("arvados.collection.Collection")
306     @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
307     def test_run(self, list_images_in_arv, mockcollection):
308         arvados_cwl.add_arv_hints()
309
310         api = mock.MagicMock()
311         api._rootDesc = get_rootDesc()
312
313         runner = arvados_cwl.ArvCwlRunner(api)
314         self.assertEqual(runner.work_api, 'jobs')
315
316         list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
317         runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
318
319         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
320         runner.ignore_docker_for_reuse = False
321         runner.num_retries = 0
322         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
323
324         tool, metadata = document_loader.resolve_ref("tests/wf/scatter2.cwl")
325         metadata["cwlVersion"] = tool["cwlVersion"]
326
327         mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118"
328
329         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
330                                          collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
331         arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names,
332                                               basedir="", make_fs_access=make_fs_access, loader=document_loader,
333                                               makeTool=runner.arv_make_tool, metadata=metadata)
334         arvtool.formatgraph = None
335         it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access)
336         it.next().run()
337         it.next().run()
338
339         with open("tests/wf/scatter2_subwf.cwl") as f:
340             subwf = StripYAMLComments(f.read())
341
342         runner.api.jobs().create.assert_called_with(
343             body=JsonDiffMatcher({
344                 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
345                 'repository': 'arvados',
346                 'script_version': 'master',
347                 'script': 'crunchrunner',
348                 'script_parameters': {
349                     'tasks': [{'task.env': {
350                         'HOME': '$(task.outdir)',
351                         'TMPDIR': '$(task.tmpdir)'},
352                                'task.vwd': {
353                                    'workflow.cwl': '$(task.keep)/99999999999999999999999999999999+118/workflow.cwl',
354                                    'cwl.input.yml': '$(task.keep)/99999999999999999999999999999999+118/cwl.input.yml'
355                                },
356                     'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'],
357                     'task.stdout': 'cwl.output.json'}]},
358                 'runtime_constraints': {
359                     'min_scratch_mb_per_node': 2048,
360                     'min_cores_per_node': 1,
361                     'docker_image': 'arvados/jobs',
362                     'min_ram_mb_per_node': 1024
363                 },
364                 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
365             filters=[['repository', '=', 'arvados'],
366                      ['script', '=', 'crunchrunner'],
367                      ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
368                      ['docker_image_locator', 'in docker', 'arvados/jobs']],
369             find_or_create=True)
370
371         mockcollection().open().__enter__().write.assert_has_calls([mock.call(subwf)])
372         mockcollection().open().__enter__().write.assert_has_calls([mock.call(
373 '''{
374   "sleeptime": 5
375 }''')])
376
377     def test_default_work_api(self):
378         arvados_cwl.add_arv_hints()
379
380         api = mock.MagicMock()
381         api._rootDesc = copy.deepcopy(get_rootDesc())
382         del api._rootDesc.get('resources')['jobs']['methods']['create']
383         runner = arvados_cwl.ArvCwlRunner(api)
384         self.assertEqual(runner.work_api, 'containers')