10165: Add test for make_output_collection
[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("draft-3")
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         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("draft-3")
76
77         tool = {
78             "inputs": [],
79             "outputs": [],
80             "hints": [{
81                 "class": "ResourceRequirement",
82                 "coresMin": 3,
83                 "ramMin": 3000,
84                 "tmpdirMin": 4000
85             }, {
86                 "class": "http://arvados.org/cwl#RuntimeConstraints",
87                 "keep_cache": 512
88             }, {
89                 "class": "http://arvados.org/cwl#APIRequirement",
90             }],
91             "baseCommand": "ls"
92         }
93         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
94         arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
95                                                  make_fs_access=make_fs_access, loader=Loader({}))
96         arvtool.formatgraph = None
97         for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
98             j.run()
99         runner.api.jobs().create.assert_called_with(
100             body={
101                 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
102                 'runtime_constraints': {},
103                 'script_parameters': {
104                     'tasks': [{
105                         'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
106                         'command': ['ls']
107                     }]
108             },
109             'script_version': 'master',
110                 'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6',
111                 'repository': 'arvados',
112                 'script': 'crunchrunner',
113                 'runtime_constraints': {
114                     'docker_image': 'arvados/jobs',
115                     'min_cores_per_node': 3,
116                     'min_ram_mb_per_node': 3000,
117                     'min_scratch_mb_per_node': 5024, # tmpdirSize + outdirSize
118                     'keep_cache_mb_per_task': 512
119                 }
120             },
121             find_or_create=True,
122             filters=[['repository', '=', 'arvados'],
123                      ['script', '=', 'crunchrunner'],
124                      ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'],
125                      ['docker_image_locator', 'in docker', 'arvados/jobs']])
126
127     @mock.patch("arvados.collection.CollectionReader")
128     def test_done(self, reader):
129         api = mock.MagicMock()
130
131         runner = mock.MagicMock()
132         runner.api = api
133         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
134         runner.num_retries = 0
135         runner.ignore_docker_for_reuse = False
136
137         reader().open.return_value = []
138         api.collections().list().execute.side_effect = ({"items": []},
139                                                         {"items": [{"manifest_text": "XYZ"}]})
140
141         arvjob = arvados_cwl.ArvadosJob(runner)
142         arvjob.name = "testjob"
143         arvjob.builder = mock.MagicMock()
144         arvjob.output_callback = mock.MagicMock()
145         arvjob.collect_outputs = mock.MagicMock()
146
147         arvjob.done({
148             "state": "Complete",
149             "output": "99999999999999999999999999999993+99",
150             "log": "99999999999999999999999999999994+99",
151             "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
152         })
153
154         api.collections().list.assert_has_calls([
155             mock.call(),
156             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
157                           ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
158                           ['name', '=', 'Output 9999999 of testjob']]),
159             mock.call().execute(num_retries=0),
160             mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999993+99']],
161                  select=['manifest_text']),
162             mock.call().execute(num_retries=0)])
163
164         api.collections().create.assert_called_with(
165             ensure_unique_name=True,
166             body={'portable_data_hash': '99999999999999999999999999999993+99',
167                   'manifest_text': 'XYZ',
168                   'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
169                   'name': 'Output 9999999 of testjob'})
170
171     @mock.patch("arvados.collection.CollectionReader")
172     def test_done_use_existing_collection(self, reader):
173         api = mock.MagicMock()
174
175         runner = mock.MagicMock()
176         runner.api = api
177         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
178         runner.num_retries = 0
179
180         reader().open.return_value = []
181         api.collections().list().execute.side_effect = ({"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},)
182
183         arvjob = arvados_cwl.ArvadosJob(runner)
184         arvjob.name = "testjob"
185         arvjob.builder = mock.MagicMock()
186         arvjob.output_callback = mock.MagicMock()
187         arvjob.collect_outputs = mock.MagicMock()
188
189         arvjob.done({
190             "state": "Complete",
191             "output": "99999999999999999999999999999993+99",
192             "log": "99999999999999999999999999999994+99",
193             "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
194         })
195
196         api.collections().list.assert_has_calls([
197             mock.call(),
198             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
199                                ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
200                                ['name', '=', 'Output 9999999 of testjob']]),
201             mock.call().execute(num_retries=0)])
202
203         self.assertFalse(api.collections().create.called)
204
205
206 class TestWorkflow(unittest.TestCase):
207     # The test passes no builder.resources
208     # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
209     @mock.patch("arvados.collection.Collection")
210     def test_run(self, mockcollection):
211         try:
212             arvados_cwl.add_arv_hints()
213
214             runner = arvados_cwl.ArvCwlRunner(mock.MagicMock())
215             runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
216             runner.ignore_docker_for_reuse = False
217             runner.num_retries = 0
218             document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
219
220             tool, metadata = document_loader.resolve_ref("tests/wf/scatter2.cwl")
221             metadata["cwlVersion"] = tool["cwlVersion"]
222
223             mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118"
224
225             make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
226             arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names,
227                                                   basedir="", make_fs_access=make_fs_access, loader=document_loader,
228                                                   makeTool=runner.arv_make_tool, metadata=metadata)
229             arvtool.formatgraph = None
230             it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access)
231             it.next().run()
232             it.next().run()
233
234             with open("tests/wf/scatter2_subwf.cwl") as f:
235                 subwf = f.read()
236
237             mockcollection().open().__enter__().write.assert_has_calls([mock.call(subwf)])
238             mockcollection().open().__enter__().write.assert_has_calls([mock.call('{sleeptime: 5}')])
239
240             runner.api.jobs().create.assert_called_with(
241                 body={
242                     'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6',
243                     'repository': 'arvados',
244                     'script_version': 'master',
245                     'script': 'crunchrunner',
246                     'script_parameters': {
247                         'tasks': [{'task.env': {
248                             'HOME': '$(task.outdir)',
249                             'TMPDIR': '$(task.tmpdir)'},
250                                    'task.vwd': {
251                                        'workflow.cwl': '$(task.keep)/99999999999999999999999999999999+118/workflow.cwl',
252                                        'cwl.input.yml': '$(task.keep)/99999999999999999999999999999999+118/cwl.input.yml'
253                                    },
254                         'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'],
255                         'task.stdout': 'cwl.output.json'}]},
256                     'runtime_constraints': {
257                         'min_scratch_mb_per_node': 2048,
258                         'min_cores_per_node': 1,
259                         'docker_image': 'arvados/jobs',
260                         'min_ram_mb_per_node': 1024
261                     },
262                     'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'},
263                 filters=[['repository', '=', 'arvados'],
264                          ['script', '=', 'crunchrunner'],
265                          ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'],
266                          ['docker_image_locator', 'in docker', 'arvados/jobs']],
267                 find_or_create=True)
268         except:
269             logging.exception("")