8784: Fix test for latest firefox.
[arvados.git] / sdk / cwl / tests / test_job.py
1 import functools
2 import json
3 import logging
4 import mock
5 import os
6 import unittest
7 import copy
8 import StringIO
9
10 import arvados
11 import arvados_cwl
12 import cwltool.process
13 from schema_salad.ref_resolver import Loader
14 from schema_salad.sourceline import cmap
15 from .mock_discovery import get_rootDesc
16 from .matcher import JsonDiffMatcher
17
18 if not os.getenv('ARVADOS_DEBUG'):
19     logging.getLogger('arvados.cwl-runner').setLevel(logging.WARN)
20     logging.getLogger('arvados.arv-run').setLevel(logging.WARN)
21
22 class TestJob(unittest.TestCase):
23
24     # The test passes no builder.resources
25     # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
26     @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
27     def test_run(self, list_images_in_arv):
28         for enable_reuse in (True, False):
29             runner = mock.MagicMock()
30             runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
31             runner.ignore_docker_for_reuse = False
32             runner.num_retries = 0
33             document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
34
35             list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
36             runner.api.collections().get().execute.return_value = {"portable_data_hash": "99999999999999999999999999999993+99"}
37             # Simulate reused job from another project so that we can check is a can_read
38             # link is added.
39             runner.api.jobs().create().execute.return_value = {
40                 'state': 'Complete' if enable_reuse else 'Queued',
41                 'owner_uuid': 'zzzzz-tpzed-yyyyyyyyyyyyyyy' if enable_reuse else 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
42                 'uuid': 'zzzzz-819sb-yyyyyyyyyyyyyyy',
43                 'output': None,
44             }
45
46             tool = cmap({
47                 "inputs": [],
48                 "outputs": [],
49                 "baseCommand": "ls",
50                 "arguments": [{"valueFrom": "$(runtime.outdir)"}]
51             })
52             make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
53                                          collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
54             arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
55                                                      basedir="", make_fs_access=make_fs_access, loader=Loader({}))
56             arvtool.formatgraph = None
57             for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
58                 j.run(enable_reuse=enable_reuse)
59                 runner.api.jobs().create.assert_called_with(
60                     body=JsonDiffMatcher({
61                         'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
62                         'runtime_constraints': {},
63                         'script_parameters': {
64                             'tasks': [{
65                                 'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
66                                 'command': ['ls', '$(task.outdir)']
67                             }],
68                         },
69                         'script_version': 'master',
70                         'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
71                         'repository': 'arvados',
72                         'script': 'crunchrunner',
73                         'runtime_constraints': {
74                             'docker_image': 'arvados/jobs',
75                             'min_cores_per_node': 1,
76                             'min_ram_mb_per_node': 1024,
77                             'min_scratch_mb_per_node': 2048 # tmpdirSize + outdirSize
78                         }
79                     }),
80                     find_or_create=enable_reuse,
81                     filters=[['repository', '=', 'arvados'],
82                              ['script', '=', 'crunchrunner'],
83                              ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
84                              ['docker_image_locator', 'in docker', 'arvados/jobs']]
85                 )
86                 if enable_reuse:
87                     runner.api.links().create.assert_called_with(
88                         body=JsonDiffMatcher({
89                             'link_class': 'permission',
90                             'name': 'can_read',
91                             "tail_uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz",
92                             "head_uuid": "zzzzz-819sb-yyyyyyyyyyyyyyy",
93                         })
94                     )
95                 else:
96                     assert not runner.api.links().create.called
97
98     # The test passes some fields in builder.resources
99     # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
100     @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
101     def test_resource_requirements(self, list_images_in_arv):
102         runner = mock.MagicMock()
103         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
104         runner.ignore_docker_for_reuse = False
105         runner.num_retries = 0
106         arvados_cwl.add_arv_hints()
107
108         list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
109         runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
110
111         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
112
113
114         tool = {
115             "inputs": [],
116             "outputs": [],
117             "hints": [{
118                 "class": "ResourceRequirement",
119                 "coresMin": 3,
120                 "ramMin": 3000,
121                 "tmpdirMin": 4000
122             }, {
123                 "class": "http://arvados.org/cwl#RuntimeConstraints",
124                 "keep_cache": 512,
125                 "outputDirType": "keep_output_dir"
126             }, {
127                 "class": "http://arvados.org/cwl#APIRequirement",
128             }],
129             "baseCommand": "ls"
130         }
131         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
132                                          collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
133         arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
134                                                  make_fs_access=make_fs_access, loader=Loader({}))
135         arvtool.formatgraph = None
136         for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
137             j.run()
138         runner.api.jobs().create.assert_called_with(
139             body=JsonDiffMatcher({
140                 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
141                 'runtime_constraints': {},
142                 'script_parameters': {
143                     'tasks': [{
144                         'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
145                         'task.keepTmpOutput': True,
146                         'command': ['ls']
147                     }]
148             },
149             'script_version': 'master',
150                 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
151                 'repository': 'arvados',
152                 'script': 'crunchrunner',
153                 'runtime_constraints': {
154                     'docker_image': 'arvados/jobs',
155                     'min_cores_per_node': 3,
156                     'min_ram_mb_per_node': 3512,     # ramMin + keep_cache
157                     'min_scratch_mb_per_node': 5024, # tmpdirSize + outdirSize
158                     'keep_cache_mb_per_task': 512
159                 }
160             }),
161             find_or_create=True,
162             filters=[['repository', '=', 'arvados'],
163                      ['script', '=', 'crunchrunner'],
164                      ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
165                      ['docker_image_locator', 'in docker', 'arvados/jobs']])
166
167     @mock.patch("arvados.collection.CollectionReader")
168     def test_done(self, reader):
169         api = mock.MagicMock()
170
171         runner = mock.MagicMock()
172         runner.api = api
173         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
174         runner.num_retries = 0
175         runner.ignore_docker_for_reuse = False
176
177         reader().open.return_value = StringIO.StringIO(
178             """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
179 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
180 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
181         """)
182         api.collections().list().execute.side_effect = ({"items": []},
183                                                         {"items": [{"manifest_text": "XYZ"}]},
184                                                         {"items": []},
185                                                         {"items": [{"manifest_text": "ABC"}]})
186
187         arvjob = arvados_cwl.ArvadosJob(runner)
188         arvjob.name = "testjob"
189         arvjob.builder = mock.MagicMock()
190         arvjob.output_callback = mock.MagicMock()
191         arvjob.collect_outputs = mock.MagicMock()
192         arvjob.collect_outputs.return_value = {"out": "stuff"}
193
194         arvjob.done({
195             "state": "Complete",
196             "output": "99999999999999999999999999999993+99",
197             "log": "99999999999999999999999999999994+99",
198             "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
199         })
200
201         api.collections().list.assert_has_calls([
202             mock.call(),
203             # Output collection check
204             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
205                           ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
206                           ['name', '=', 'Output 9999999 of testjob']]),
207             mock.call().execute(num_retries=0),
208             mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999993+99']],
209                  select=['manifest_text']),
210             mock.call().execute(num_retries=0),
211             # Log collection's turn
212             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
213                           ['portable_data_hash', '=', '99999999999999999999999999999994+99'],
214                           ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]),
215             mock.call().execute(num_retries=0),
216             mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999994+99']],
217                  select=['manifest_text']),
218             mock.call().execute(num_retries=0)])
219
220         api.collections().create.assert_has_calls([
221             mock.call(ensure_unique_name=True,
222                       body={'portable_data_hash': '99999999999999999999999999999993+99',
223                             'manifest_text': 'XYZ',
224                             'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
225                             'name': 'Output 9999999 of testjob'}),
226             mock.call().execute(num_retries=0),
227             mock.call(ensure_unique_name=True,
228                       body={'portable_data_hash': '99999999999999999999999999999994+99',
229                             'manifest_text': 'ABC',
230                             'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
231                             'name': 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
232             mock.call().execute(num_retries=0),
233         ])
234
235         arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
236
237     @mock.patch("arvados.collection.CollectionReader")
238     def test_done_use_existing_collection(self, reader):
239         api = mock.MagicMock()
240
241         runner = mock.MagicMock()
242         runner.api = api
243         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
244         runner.num_retries = 0
245
246         reader().open.return_value = StringIO.StringIO(
247             """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
248 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
249 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
250         """)
251
252         api.collections().list().execute.side_effect = (
253             {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},
254             {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},
255         )
256
257         arvjob = arvados_cwl.ArvadosJob(runner)
258         arvjob.name = "testjob"
259         arvjob.builder = mock.MagicMock()
260         arvjob.output_callback = mock.MagicMock()
261         arvjob.collect_outputs = mock.MagicMock()
262         arvjob.collect_outputs.return_value = {"out": "stuff"}
263
264         arvjob.done({
265             "state": "Complete",
266             "output": "99999999999999999999999999999993+99",
267             "log": "99999999999999999999999999999994+99",
268             "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
269         })
270
271         api.collections().list.assert_has_calls([
272             mock.call(),
273             # Output collection
274             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
275                                ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
276                                ['name', '=', 'Output 9999999 of testjob']]),
277             mock.call().execute(num_retries=0),
278             # Log collection
279             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
280                                ['portable_data_hash', '=', '99999999999999999999999999999994+99'],
281                                ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]),
282             mock.call().execute(num_retries=0)
283         ])
284
285         self.assertFalse(api.collections().create.called)
286
287         arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
288
289
290 class TestWorkflow(unittest.TestCase):
291     # The test passes no builder.resources
292     # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
293     @mock.patch("arvados.collection.Collection")
294     @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
295     def test_run(self, list_images_in_arv, mockcollection):
296         arvados_cwl.add_arv_hints()
297
298         api = mock.MagicMock()
299         api._rootDesc = get_rootDesc()
300
301         runner = arvados_cwl.ArvCwlRunner(api)
302         self.assertEqual(runner.work_api, 'jobs')
303
304         list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
305         runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
306
307         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
308         runner.ignore_docker_for_reuse = False
309         runner.num_retries = 0
310         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
311
312         tool, metadata = document_loader.resolve_ref("tests/wf/scatter2.cwl")
313         metadata["cwlVersion"] = tool["cwlVersion"]
314
315         mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118"
316
317         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
318                                          collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
319         arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names,
320                                               basedir="", make_fs_access=make_fs_access, loader=document_loader,
321                                               makeTool=runner.arv_make_tool, metadata=metadata)
322         arvtool.formatgraph = None
323         it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access)
324         it.next().run()
325         it.next().run()
326
327         with open("tests/wf/scatter2_subwf.cwl") as f:
328             subwf = f.read()
329
330         runner.api.jobs().create.assert_called_with(
331             body=JsonDiffMatcher({
332                 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
333                 'repository': 'arvados',
334                 'script_version': 'master',
335                 'script': 'crunchrunner',
336                 'script_parameters': {
337                     'tasks': [{'task.env': {
338                         'HOME': '$(task.outdir)',
339                         'TMPDIR': '$(task.tmpdir)'},
340                                'task.vwd': {
341                                    'workflow.cwl': '$(task.keep)/99999999999999999999999999999999+118/workflow.cwl',
342                                    'cwl.input.yml': '$(task.keep)/99999999999999999999999999999999+118/cwl.input.yml'
343                                },
344                     'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'],
345                     'task.stdout': 'cwl.output.json'}]},
346                 'runtime_constraints': {
347                     'min_scratch_mb_per_node': 2048,
348                     'min_cores_per_node': 1,
349                     'docker_image': 'arvados/jobs',
350                     'min_ram_mb_per_node': 1024
351                 },
352                 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
353             filters=[['repository', '=', 'arvados'],
354                      ['script', '=', 'crunchrunner'],
355                      ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
356                      ['docker_image_locator', 'in docker', 'arvados/jobs']],
357             find_or_create=True)
358
359         mockcollection().open().__enter__().write.assert_has_calls([mock.call(subwf)])
360         mockcollection().open().__enter__().write.assert_has_calls([mock.call(
361 '''{
362   "sleeptime": 5
363 }''')])
364
365     def test_default_work_api(self):
366         arvados_cwl.add_arv_hints()
367
368         api = mock.MagicMock()
369         api._rootDesc = copy.deepcopy(get_rootDesc())
370         del api._rootDesc.get('resources')['jobs']['methods']['create']
371         runner = arvados_cwl.ArvCwlRunner(api)
372         self.assertEqual(runner.work_api, 'containers')