Merge branch '9397-cwl-initialworkdir-crunchv2' closes #9397
[arvados.git] / sdk / cwl / tests / test_container.py
1 import arvados_cwl
2 from arvados_cwl.arvdocker import arv_docker_clear_cache
3 import logging
4 import mock
5 import unittest
6 import os
7 import functools
8 import cwltool.process
9 from schema_salad.ref_resolver import Loader
10 from schema_salad.sourceline import cmap
11
12 from .matcher import JsonDiffMatcher
13
14 if not os.getenv('ARVADOS_DEBUG'):
15     logging.getLogger('arvados.cwl-runner').setLevel(logging.WARN)
16     logging.getLogger('arvados.arv-run').setLevel(logging.WARN)
17
18
19 class TestContainer(unittest.TestCase):
20
21     # The test passes no builder.resources
22     # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
23     @mock.patch("arvados.commands.keepdocker.list_images_in_arv")
24     def test_run(self, keepdocker):
25         for enable_reuse in (True, False):
26             arv_docker_clear_cache()
27
28             runner = mock.MagicMock()
29             runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
30             runner.ignore_docker_for_reuse = False
31
32             keepdocker.return_value = [("zzzzz-4zz18-zzzzzzzzzzzzzz3", "")]
33             runner.api.collections().get().execute.return_value = {
34                 "portable_data_hash": "99999999999999999999999999999993+99"}
35
36             document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
37
38             tool = cmap({
39                 "inputs": [],
40                 "outputs": [],
41                 "baseCommand": "ls",
42                 "arguments": [{"valueFrom": "$(runtime.outdir)"}]
43             })
44             make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
45             arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="containers", avsc_names=avsc_names,
46                                                      basedir="", make_fs_access=make_fs_access, loader=Loader({}))
47             arvtool.formatgraph = None
48             for j in arvtool.job({}, mock.MagicMock(), basedir="", name="test_run_"+str(enable_reuse),
49                                  make_fs_access=make_fs_access, tmpdir="/tmp"):
50                 j.run(enable_reuse=enable_reuse)
51                 runner.api.container_requests().create.assert_called_with(
52                     body=JsonDiffMatcher({
53                         'environment': {
54                             'HOME': '/var/spool/cwl',
55                             'TMPDIR': '/tmp'
56                         },
57                         'name': 'test_run_'+str(enable_reuse),
58                         'runtime_constraints': {
59                             'vcpus': 1,
60                             'ram': 1073741824
61                         },
62                         'use_existing': enable_reuse,
63                         'priority': 1,
64                         'mounts': {
65                             '/var/spool/cwl': {'kind': 'tmp'}
66                         },
67                         'state': 'Committed',
68                         'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
69                         'output_path': '/var/spool/cwl',
70                         'container_image': '99999999999999999999999999999993+99',
71                         'command': ['ls', '/var/spool/cwl'],
72                         'cwd': '/var/spool/cwl',
73                         'scheduling_parameters': {},
74                         'properties': {},
75                     }))
76
77     # The test passes some fields in builder.resources
78     # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
79     @mock.patch("arvados.commands.keepdocker.list_images_in_arv")
80     def test_resource_requirements(self, keepdocker):
81         arv_docker_clear_cache()
82         runner = mock.MagicMock()
83         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
84         runner.ignore_docker_for_reuse = False
85         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
86
87         keepdocker.return_value = [("zzzzz-4zz18-zzzzzzzzzzzzzz3", "")]
88         runner.api.collections().get().execute.return_value = {
89             "portable_data_hash": "99999999999999999999999999999993+99"}
90
91         tool = cmap({
92             "inputs": [],
93             "outputs": [],
94             "hints": [{
95                 "class": "ResourceRequirement",
96                 "coresMin": 3,
97                 "ramMin": 3000,
98                 "tmpdirMin": 4000
99             }, {
100                 "class": "http://arvados.org/cwl#RuntimeConstraints",
101                 "keep_cache": 512
102             }, {
103                 "class": "http://arvados.org/cwl#APIRequirement",
104             }, {
105                 "class": "http://arvados.org/cwl#PartitionRequirement",
106                 "partition": "blurb"
107             }],
108             "baseCommand": "ls"
109         })
110         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
111         arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="containers",
112                                                  avsc_names=avsc_names, make_fs_access=make_fs_access,
113                                                  loader=Loader({}))
114         arvtool.formatgraph = None
115         for j in arvtool.job({}, mock.MagicMock(), basedir="", name="test_resource_requirements",
116                              make_fs_access=make_fs_access, tmpdir="/tmp"):
117             j.run()
118
119         call_args, call_kwargs = runner.api.container_requests().create.call_args
120
121         call_body_expected = {
122                 'environment': {
123                     'HOME': '/var/spool/cwl',
124                     'TMPDIR': '/tmp'
125                 },
126                 'name': 'test_resource_requirements',
127                 'runtime_constraints': {
128                     'vcpus': 3,
129                     'ram': 3145728000,
130                     'keep_cache_ram': 536870912,
131                     'API': True
132                 },
133                 'use_existing': True,
134                 'priority': 1,
135                 'mounts': {
136                     '/var/spool/cwl': {'kind': 'tmp'}
137                 },
138                 'state': 'Committed',
139                 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
140                 'output_path': '/var/spool/cwl',
141                 'container_image': '99999999999999999999999999999993+99',
142                 'command': ['ls'],
143                 'cwd': '/var/spool/cwl',
144                 'scheduling_parameters': {
145                     'partitions': ['blurb']
146                 },
147                 'properties': {}
148         }
149
150         call_body = call_kwargs.get('body', None)
151         self.assertNotEqual(None, call_body)
152         for key in call_body:
153             self.assertEqual(call_body_expected.get(key), call_body.get(key))
154
155
156     # The test passes some fields in builder.resources
157     # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
158     @mock.patch("arvados.commands.keepdocker.list_images_in_arv")
159     @mock.patch("arvados.collection.Collection")
160     def test_initial_work_dir(self, collection_mock, keepdocker):
161         arv_docker_clear_cache()
162         runner = mock.MagicMock()
163         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
164         runner.ignore_docker_for_reuse = False
165         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
166
167         keepdocker.return_value = [("zzzzz-4zz18-zzzzzzzzzzzzzz3", "")]
168         runner.api.collections().get().execute.return_value = {
169             "portable_data_hash": "99999999999999999999999999999993+99"}
170
171         sourcemock = mock.MagicMock()
172         def get_collection_mock(p):
173             if "/" in p:
174                 return (sourcemock, p.split("/", 1)[1])
175             else:
176                 return (sourcemock, "")
177         runner.fs_access.get_collection.side_effect = get_collection_mock
178
179         vwdmock = mock.MagicMock()
180         collection_mock.return_value = vwdmock
181         vwdmock.portable_data_hash.return_value = "99999999999999999999999999999996+99"
182
183         tool = cmap({
184             "inputs": [],
185             "outputs": [],
186             "hints": [{
187                 "class": "InitialWorkDirRequirement",
188                 "listing": [{
189                     "class": "File",
190                     "basename": "foo",
191                     "location": "keep:99999999999999999999999999999995+99/bar"
192                 },
193                 {
194                     "class": "Directory",
195                     "basename": "foo2",
196                     "location": "keep:99999999999999999999999999999995+99"
197                 },
198                 {
199                     "class": "File",
200                     "basename": "filename",
201                     "location": "keep:99999999999999999999999999999995+99/baz/filename"
202                 },
203                 {
204                     "class": "Directory",
205                     "basename": "subdir",
206                     "location": "keep:99999999999999999999999999999995+99/subdir"
207                 }                        ]
208             }],
209             "baseCommand": "ls"
210         })
211         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
212         arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="containers",
213                                                  avsc_names=avsc_names, make_fs_access=make_fs_access,
214                                                  loader=Loader({}))
215         arvtool.formatgraph = None
216         for j in arvtool.job({}, mock.MagicMock(), basedir="", name="test_initial_work_dir",
217                              make_fs_access=make_fs_access, tmpdir="/tmp"):
218             j.run()
219
220         call_args, call_kwargs = runner.api.container_requests().create.call_args
221
222         vwdmock.copy.assert_has_calls([mock.call('bar', 'foo', source_collection=sourcemock)])
223         vwdmock.copy.assert_has_calls([mock.call('', 'foo2', source_collection=sourcemock)])
224         vwdmock.copy.assert_has_calls([mock.call('baz/filename', 'filename', source_collection=sourcemock)])
225         vwdmock.copy.assert_has_calls([mock.call('subdir', 'subdir', source_collection=sourcemock)])
226
227         call_body_expected = {
228                 'environment': {
229                     'HOME': '/var/spool/cwl',
230                     'TMPDIR': '/tmp'
231                 },
232                 'name': 'test_initial_work_dir',
233                 'runtime_constraints': {
234                     'vcpus': 1,
235                     'ram': 1073741824
236                 },
237                 'use_existing': True,
238                 'priority': 1,
239                 'mounts': {
240                     '/var/spool/cwl': {'kind': 'tmp'},
241                     '/var/spool/cwl/foo': {
242                         'kind': 'collection',
243                         'path': 'foo',
244                         'portable_data_hash': '99999999999999999999999999999996+99'
245                     },
246                     '/var/spool/cwl/foo2': {
247                         'kind': 'collection',
248                         'path': 'foo2',
249                         'portable_data_hash': '99999999999999999999999999999996+99'
250                     },
251                     '/var/spool/cwl/filename': {
252                         'kind': 'collection',
253                         'path': 'filename',
254                         'portable_data_hash': '99999999999999999999999999999996+99'
255                     },
256                     '/var/spool/cwl/subdir': {
257                         'kind': 'collection',
258                         'path': 'subdir',
259                         'portable_data_hash': '99999999999999999999999999999996+99'
260                     }
261                 },
262                 'state': 'Committed',
263                 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
264                 'output_path': '/var/spool/cwl',
265                 'container_image': '99999999999999999999999999999993+99',
266                 'command': ['ls'],
267                 'cwd': '/var/spool/cwl',
268                 'scheduling_parameters': {
269                 },
270                 'properties': {}
271         }
272
273         call_body = call_kwargs.get('body', None)
274         self.assertNotEqual(None, call_body)
275         for key in call_body:
276             self.assertEqual(call_body_expected.get(key), call_body.get(key))
277
278     @mock.patch("arvados.collection.Collection")
279     def test_done(self, col):
280         api = mock.MagicMock()
281
282         runner = mock.MagicMock()
283         runner.api = api
284         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
285         runner.num_retries = 0
286         runner.ignore_docker_for_reuse = False
287
288         runner.api.containers().get().execute.return_value = {"state":"Complete",
289                                                               "output": "abc+123",
290                                                               "exit_code": 0}
291
292         col().open.return_value = []
293
294         arvjob = arvados_cwl.ArvadosContainer(runner)
295         arvjob.name = "testjob"
296         arvjob.builder = mock.MagicMock()
297         arvjob.output_callback = mock.MagicMock()
298         arvjob.collect_outputs = mock.MagicMock()
299         arvjob.successCodes = [0]
300         arvjob.outdir = "/var/spool/cwl"
301
302         arvjob.collect_outputs.return_value = {"out": "stuff"}
303
304         arvjob.done({
305             "state": "Final",
306             "log_uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz1",
307             "output_uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2",
308             "uuid": "zzzzz-xvhdp-zzzzzzzzzzzzzzz",
309             "container_uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
310         })
311
312         self.assertFalse(api.collections().create.called)
313
314         arvjob.collect_outputs.assert_called_with("keep:abc+123")
315         arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")