Merge branch 'master' into 9307-cwl-use-tmp-output
[arvados.git] / sdk / cwl / tests / test_job.py
1 import logging
2 import mock
3 import unittest
4 import os
5 import functools
6 import json
7
8 import arvados_cwl
9 import cwltool.process
10 from schema_salad.ref_resolver import Loader
11
12 if not os.getenv('ARVADOS_DEBUG'):
13     logging.getLogger('arvados.cwl-runner').setLevel(logging.WARN)
14     logging.getLogger('arvados.arv-run').setLevel(logging.WARN)
15
16
17 class TestJob(unittest.TestCase):
18
19     # The test passes no builder.resources
20     # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
21     def test_run(self):
22         runner = mock.MagicMock()
23         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
24         runner.ignore_docker_for_reuse = False
25         runner.num_retries = 0
26         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
27
28         tool = {
29             "inputs": [],
30             "outputs": [],
31             "baseCommand": "ls",
32             "arguments": [{"valueFrom": "$(runtime.outdir)"}]
33         }
34         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
35         arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
36                                                  basedir="", make_fs_access=make_fs_access, loader=Loader({}))
37         arvtool.formatgraph = None
38         for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
39             j.run()
40             runner.api.jobs().create.assert_called_with(
41                 body={
42                     'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
43                     'runtime_constraints': {},
44                     'script_parameters': {
45                         'tasks': [{
46                             'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
47                             'command': ['ls', '$(task.outdir)']
48                         }],
49                     },
50                     'script_version': 'master',
51                     'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6',
52                     'repository': 'arvados',
53                     'script': 'crunchrunner',
54                     'runtime_constraints': {
55                         'docker_image': 'arvados/jobs',
56                         'min_cores_per_node': 1,
57                         'min_ram_mb_per_node': 1024,
58                         'min_scratch_mb_per_node': 2048 # tmpdirSize + outdirSize
59                     }
60                 },
61                 find_or_create=True,
62                 filters=[['repository', '=', 'arvados'],
63                          ['script', '=', 'crunchrunner'],
64                          ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'],
65                          ['docker_image_locator', 'in docker', 'arvados/jobs']]
66             )
67
68     # The test passes some fields in builder.resources
69     # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
70     def test_resource_requirements(self):
71         runner = mock.MagicMock()
72         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
73         runner.ignore_docker_for_reuse = False
74         runner.num_retries = 0
75         arvados_cwl.add_arv_hints()
76
77         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
78
79
80         tool = {
81             "inputs": [],
82             "outputs": [],
83             "hints": [{
84                 "class": "ResourceRequirement",
85                 "coresMin": 3,
86                 "ramMin": 3000,
87                 "tmpdirMin": 4000
88             }, {
89                 "class": "http://arvados.org/cwl#RuntimeConstraints",
90                 "keep_cache": 512,
91                 "outputDirType": "keep_output_dir"
92             }, {
93                 "class": "http://arvados.org/cwl#APIRequirement",
94             }],
95             "baseCommand": "ls"
96         }
97         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
98         arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
99                                                  make_fs_access=make_fs_access, loader=Loader({}))
100         arvtool.formatgraph = None
101         for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
102             j.run()
103         runner.api.jobs().create.assert_called_with(
104             body={
105                 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
106                 'runtime_constraints': {},
107                 'script_parameters': {
108                     'tasks': [{
109                         'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
110                         'task.keepTmpOutput': True,
111                         'command': ['ls']
112                     }]
113             },
114             'script_version': 'master',
115                 'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6',
116                 'repository': 'arvados',
117                 'script': 'crunchrunner',
118                 'runtime_constraints': {
119                     'docker_image': 'arvados/jobs',
120                     'min_cores_per_node': 3,
121                     'min_ram_mb_per_node': 3000,
122                     'min_scratch_mb_per_node': 5024, # tmpdirSize + outdirSize
123                     'keep_cache_mb_per_task': 512
124                 }
125             },
126             find_or_create=True,
127             filters=[['repository', '=', 'arvados'],
128                      ['script', '=', 'crunchrunner'],
129                      ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'],
130                      ['docker_image_locator', 'in docker', 'arvados/jobs']])
131
132     @mock.patch("arvados.collection.CollectionReader")
133     def test_done(self, reader):
134         api = mock.MagicMock()
135
136         runner = mock.MagicMock()
137         runner.api = api
138         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
139         runner.num_retries = 0
140         runner.ignore_docker_for_reuse = False
141
142         reader().open.return_value = []
143         api.collections().list().execute.side_effect = ({"items": []},
144                                                         {"items": [{"manifest_text": "XYZ"}]})
145
146         arvjob = arvados_cwl.ArvadosJob(runner)
147         arvjob.name = "testjob"
148         arvjob.builder = mock.MagicMock()
149         arvjob.output_callback = mock.MagicMock()
150         arvjob.collect_outputs = mock.MagicMock()
151
152         arvjob.done({
153             "state": "Complete",
154             "output": "99999999999999999999999999999993+99",
155             "log": "99999999999999999999999999999994+99",
156             "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
157         })
158
159         api.collections().list.assert_has_calls([
160             mock.call(),
161             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
162                           ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
163                           ['name', '=', 'Output 9999999 of testjob']]),
164             mock.call().execute(num_retries=0),
165             mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999993+99']],
166                  select=['manifest_text']),
167             mock.call().execute(num_retries=0)])
168
169         api.collections().create.assert_called_with(
170             ensure_unique_name=True,
171             body={'portable_data_hash': '99999999999999999999999999999993+99',
172                   'manifest_text': 'XYZ',
173                   'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
174                   'name': 'Output 9999999 of testjob'})
175
176     @mock.patch("arvados.collection.CollectionReader")
177     def test_done_use_existing_collection(self, reader):
178         api = mock.MagicMock()
179
180         runner = mock.MagicMock()
181         runner.api = api
182         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
183         runner.num_retries = 0
184
185         reader().open.return_value = []
186         api.collections().list().execute.side_effect = ({"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},)
187
188         arvjob = arvados_cwl.ArvadosJob(runner)
189         arvjob.name = "testjob"
190         arvjob.builder = mock.MagicMock()
191         arvjob.output_callback = mock.MagicMock()
192         arvjob.collect_outputs = mock.MagicMock()
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             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
204                                ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
205                                ['name', '=', 'Output 9999999 of testjob']]),
206             mock.call().execute(num_retries=0)])
207
208         self.assertFalse(api.collections().create.called)
209
210
211 class TestWorkflow(unittest.TestCase):
212     # The test passes no builder.resources
213     # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
214     @mock.patch("arvados.collection.Collection")
215     def test_run(self, mockcollection):
216         try:
217             arvados_cwl.add_arv_hints()
218
219             runner = arvados_cwl.ArvCwlRunner(mock.MagicMock())
220             runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
221             runner.ignore_docker_for_reuse = False
222             runner.num_retries = 0
223             document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
224
225             tool, metadata = document_loader.resolve_ref("tests/wf/scatter2.cwl")
226             metadata["cwlVersion"] = tool["cwlVersion"]
227
228             mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118"
229
230             make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
231             arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names,
232                                                   basedir="", make_fs_access=make_fs_access, loader=document_loader,
233                                                   makeTool=runner.arv_make_tool, metadata=metadata)
234             arvtool.formatgraph = None
235             it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access)
236             it.next().run()
237             it.next().run()
238
239             with open("tests/wf/scatter2_subwf.cwl") as f:
240                 subwf = f.read()
241
242             mockcollection().open().__enter__().write.assert_has_calls([mock.call(subwf)])
243             mockcollection().open().__enter__().write.assert_has_calls([mock.call('{sleeptime: 5}')])
244
245             runner.api.jobs().create.assert_called_with(
246                 body={
247                     'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6',
248                     'repository': 'arvados',
249                     'script_version': 'master',
250                     'script': 'crunchrunner',
251                     'script_parameters': {
252                         'tasks': [{'task.env': {
253                             'HOME': '$(task.outdir)',
254                             'TMPDIR': '$(task.tmpdir)'},
255                                    'task.vwd': {
256                                        'workflow.cwl': '$(task.keep)/99999999999999999999999999999999+118/workflow.cwl',
257                                        'cwl.input.yml': '$(task.keep)/99999999999999999999999999999999+118/cwl.input.yml'
258                                    },
259                         'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'],
260                         'task.stdout': 'cwl.output.json'}]},
261                     'runtime_constraints': {
262                         'min_scratch_mb_per_node': 2048,
263                         'min_cores_per_node': 1,
264                         'docker_image': 'arvados/jobs',
265                         'min_ram_mb_per_node': 1024
266                     },
267                     'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'},
268                 filters=[['repository', '=', 'arvados'],
269                          ['script', '=', 'crunchrunner'],
270                          ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'],
271                          ['docker_image_locator', 'in docker', 'arvados/jobs']],
272                 find_or_create=True)
273         except:
274             logging.exception("")