Merge branch 'master' of git.curoverse.com:arvados into 11876-r-sdk
[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": "$(runtime.outdir)"}],
56                 "id": "#"
57             })
58             make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
59                                          collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
60             arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
61                                                      basedir="", make_fs_access=make_fs_access, loader=Loader({}))
62             arvtool.formatgraph = None
63             for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
64                 j.run(enable_reuse=enable_reuse)
65                 runner.api.jobs().create.assert_called_with(
66                     body=JsonDiffMatcher({
67                         'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
68                         'runtime_constraints': {},
69                         'script_parameters': {
70                             'tasks': [{
71                                 'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
72                                 'command': ['ls', '$(task.outdir)']
73                             }],
74                         },
75                         'script_version': 'master',
76                         'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
77                         'repository': 'arvados',
78                         'script': 'crunchrunner',
79                         'runtime_constraints': {
80                             'docker_image': 'arvados/jobs',
81                             'min_cores_per_node': 1,
82                             'min_ram_mb_per_node': 1024,
83                             'min_scratch_mb_per_node': 2048 # tmpdirSize + outdirSize
84                         }
85                     }),
86                     find_or_create=enable_reuse,
87                     filters=[['repository', '=', 'arvados'],
88                              ['script', '=', 'crunchrunner'],
89                              ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
90                              ['docker_image_locator', 'in docker', 'arvados/jobs']]
91                 )
92                 if enable_reuse:
93                     runner.api.links().create.assert_called_with(
94                         body=JsonDiffMatcher({
95                             'link_class': 'permission',
96                             'name': 'can_read',
97                             "tail_uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz",
98                             "head_uuid": "zzzzz-819sb-yyyyyyyyyyyyyyy",
99                         })
100                     )
101                     # Simulate an API excepction when trying to create a
102                     # sharing link on the job
103                     runner.api.links().create.side_effect = ApiError(
104                         mock.MagicMock(return_value={'status': 403}),
105                         'Permission denied')
106                     j.run(enable_reuse=enable_reuse)
107                     j.output_callback.assert_called_with({}, 'success')
108                 else:
109                     assert not runner.api.links().create.called
110
111     # The test passes some fields in builder.resources
112     # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
113     @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
114     def test_resource_requirements(self, list_images_in_arv):
115         runner = mock.MagicMock()
116         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
117         runner.ignore_docker_for_reuse = False
118         runner.num_retries = 0
119         arvados_cwl.add_arv_hints()
120
121         list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
122         runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
123
124         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
125
126
127         tool = {
128             "inputs": [],
129             "outputs": [],
130             "hints": [{
131                 "class": "ResourceRequirement",
132                 "coresMin": 3,
133                 "ramMin": 3000,
134                 "tmpdirMin": 4000
135             }, {
136                 "class": "http://arvados.org/cwl#RuntimeConstraints",
137                 "keep_cache": 512,
138                 "outputDirType": "keep_output_dir"
139             }, {
140                 "class": "http://arvados.org/cwl#APIRequirement",
141             },
142             {
143                 "class": "http://arvados.org/cwl#ReuseRequirement",
144                 "enableReuse": False
145             }],
146             "baseCommand": "ls",
147             "id": "#"
148         }
149         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
150                                          collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
151         arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
152                                                  make_fs_access=make_fs_access, loader=Loader({}))
153         arvtool.formatgraph = None
154         for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
155             j.run(enable_reuse=True)
156         runner.api.jobs().create.assert_called_with(
157             body=JsonDiffMatcher({
158                 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
159                 'runtime_constraints': {},
160                 'script_parameters': {
161                     'tasks': [{
162                         'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
163                         'task.keepTmpOutput': True,
164                         'command': ['ls']
165                     }]
166             },
167             'script_version': 'master',
168                 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
169                 'repository': 'arvados',
170                 'script': 'crunchrunner',
171                 'runtime_constraints': {
172                     'docker_image': 'arvados/jobs',
173                     'min_cores_per_node': 3,
174                     'min_ram_mb_per_node': 3512,     # ramMin + keep_cache
175                     'min_scratch_mb_per_node': 5024, # tmpdirSize + outdirSize
176                     'keep_cache_mb_per_task': 512
177                 }
178             }),
179             find_or_create=False,
180             filters=[['repository', '=', 'arvados'],
181                      ['script', '=', 'crunchrunner'],
182                      ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
183                      ['docker_image_locator', 'in docker', 'arvados/jobs']])
184
185     @mock.patch("arvados.collection.CollectionReader")
186     def test_done(self, reader):
187         api = mock.MagicMock()
188
189         runner = mock.MagicMock()
190         runner.api = api
191         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
192         runner.num_retries = 0
193         runner.ignore_docker_for_reuse = False
194
195         reader().open.return_value = StringIO.StringIO(
196             """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
197 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
198 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
199         """)
200         api.collections().list().execute.side_effect = ({"items": []},
201                                                         {"items": [{"manifest_text": "XYZ"}]},
202                                                         {"items": []},
203                                                         {"items": [{"manifest_text": "ABC"}]})
204
205         arvjob = arvados_cwl.ArvadosJob(runner)
206         arvjob.name = "testjob"
207         arvjob.builder = mock.MagicMock()
208         arvjob.output_callback = mock.MagicMock()
209         arvjob.collect_outputs = mock.MagicMock()
210         arvjob.collect_outputs.return_value = {"out": "stuff"}
211
212         arvjob.done({
213             "state": "Complete",
214             "output": "99999999999999999999999999999993+99",
215             "log": "99999999999999999999999999999994+99",
216             "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
217         })
218
219         api.collections().list.assert_has_calls([
220             mock.call(),
221             # Output collection check
222             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
223                           ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
224                           ['name', '=', 'Output 9999999 of testjob']]),
225             mock.call().execute(num_retries=0),
226             mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999993+99']],
227                  select=['manifest_text']),
228             mock.call().execute(num_retries=0),
229             # Log collection's turn
230             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
231                           ['portable_data_hash', '=', '99999999999999999999999999999994+99'],
232                           ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]),
233             mock.call().execute(num_retries=0),
234             mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999994+99']],
235                  select=['manifest_text']),
236             mock.call().execute(num_retries=0)])
237
238         api.collections().create.assert_has_calls([
239             mock.call(ensure_unique_name=True,
240                       body={'portable_data_hash': '99999999999999999999999999999993+99',
241                             'manifest_text': 'XYZ',
242                             'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
243                             'name': 'Output 9999999 of testjob'}),
244             mock.call().execute(num_retries=0),
245             mock.call(ensure_unique_name=True,
246                       body={'portable_data_hash': '99999999999999999999999999999994+99',
247                             'manifest_text': 'ABC',
248                             'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
249                             'name': 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
250             mock.call().execute(num_retries=0),
251         ])
252
253         arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
254
255     @mock.patch("arvados.collection.CollectionReader")
256     def test_done_use_existing_collection(self, reader):
257         api = mock.MagicMock()
258
259         runner = mock.MagicMock()
260         runner.api = api
261         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
262         runner.num_retries = 0
263
264         reader().open.return_value = StringIO.StringIO(
265             """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
266 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
267 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
268         """)
269
270         api.collections().list().execute.side_effect = (
271             {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},
272             {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},
273         )
274
275         arvjob = arvados_cwl.ArvadosJob(runner)
276         arvjob.name = "testjob"
277         arvjob.builder = mock.MagicMock()
278         arvjob.output_callback = mock.MagicMock()
279         arvjob.collect_outputs = mock.MagicMock()
280         arvjob.collect_outputs.return_value = {"out": "stuff"}
281
282         arvjob.done({
283             "state": "Complete",
284             "output": "99999999999999999999999999999993+99",
285             "log": "99999999999999999999999999999994+99",
286             "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
287         })
288
289         api.collections().list.assert_has_calls([
290             mock.call(),
291             # Output collection
292             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
293                                ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
294                                ['name', '=', 'Output 9999999 of testjob']]),
295             mock.call().execute(num_retries=0),
296             # Log collection
297             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
298                                ['portable_data_hash', '=', '99999999999999999999999999999994+99'],
299                                ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]),
300             mock.call().execute(num_retries=0)
301         ])
302
303         self.assertFalse(api.collections().create.called)
304
305         arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
306
307
308 class TestWorkflow(unittest.TestCase):
309     # The test passes no builder.resources
310     # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
311     @mock.patch("arvados.collection.CollectionReader")
312     @mock.patch("arvados.collection.Collection")
313     @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
314     def test_run(self, list_images_in_arv, mockcollection, mockcollectionreader):
315         arvados_cwl.add_arv_hints()
316
317         api = mock.MagicMock()
318         api._rootDesc = get_rootDesc()
319
320         runner = arvados_cwl.ArvCwlRunner(api)
321         self.assertEqual(runner.work_api, 'jobs')
322
323         list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
324         runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
325         runner.api.collections().list().execute.return_vaulue = {"items": [{"portable_data_hash": "99999999999999999999999999999993+99"}]}
326
327         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
328         runner.ignore_docker_for_reuse = False
329         runner.num_retries = 0
330         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
331
332         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
333                                          collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
334         document_loader.fetcher_constructor = functools.partial(arvados_cwl.CollectionFetcher, api_client=api, fs_access=make_fs_access(""))
335         document_loader.fetcher = document_loader.fetcher_constructor(document_loader.cache, document_loader.session)
336         document_loader.fetch_text = document_loader.fetcher.fetch_text
337         document_loader.check_exists = document_loader.fetcher.check_exists
338
339         tool, metadata = document_loader.resolve_ref("tests/wf/scatter2.cwl")
340         metadata["cwlVersion"] = tool["cwlVersion"]
341
342         mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118"
343
344         arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names,
345                                               basedir="", make_fs_access=make_fs_access, loader=document_loader,
346                                               makeTool=runner.arv_make_tool, metadata=metadata)
347         arvtool.formatgraph = None
348         it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access)
349         it.next().run()
350         it.next().run()
351
352         with open("tests/wf/scatter2_subwf.cwl") as f:
353             subwf = StripYAMLComments(f.read())
354
355         runner.api.jobs().create.assert_called_with(
356             body=JsonDiffMatcher({
357                 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
358                 'repository': 'arvados',
359                 'script_version': 'master',
360                 'script': 'crunchrunner',
361                 'script_parameters': {
362                     'tasks': [{'task.env': {
363                         'HOME': '$(task.outdir)',
364                         'TMPDIR': '$(task.tmpdir)'},
365                                'task.vwd': {
366                                    'workflow.cwl': '$(task.keep)/99999999999999999999999999999999+118/workflow.cwl',
367                                    'cwl.input.yml': '$(task.keep)/99999999999999999999999999999999+118/cwl.input.yml'
368                                },
369                     'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'],
370                     'task.stdout': 'cwl.output.json'}]},
371                 'runtime_constraints': {
372                     'min_scratch_mb_per_node': 2048,
373                     'min_cores_per_node': 1,
374                     'docker_image': 'arvados/jobs',
375                     'min_ram_mb_per_node': 1024
376                 },
377                 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
378             filters=[['repository', '=', 'arvados'],
379                      ['script', '=', 'crunchrunner'],
380                      ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
381                      ['docker_image_locator', 'in docker', 'arvados/jobs']],
382             find_or_create=True)
383
384         mockcollection().open().__enter__().write.assert_has_calls([mock.call(subwf)])
385         mockcollection().open().__enter__().write.assert_has_calls([mock.call(
386 '''{
387   "fileblub": {
388     "basename": "token.txt",
389     "class": "File",
390     "location": "/keep/99999999999999999999999999999999+118/token.txt"
391   },
392   "sleeptime": 5
393 }''')])
394
395     def test_default_work_api(self):
396         arvados_cwl.add_arv_hints()
397
398         api = mock.MagicMock()
399         api._rootDesc = copy.deepcopy(get_rootDesc())
400         del api._rootDesc.get('resources')['jobs']['methods']['create']
401         runner = arvados_cwl.ArvCwlRunner(api)
402         self.assertEqual(runner.work_api, 'containers')