Merge branch '2411-check-copyright'
[arvados.git] / sdk / cwl / tests / test_container.py
1 # Copyright (C) The Arvados Authors. All rights reserved.
2 #
3 # SPDX-License-Identifier: Apache-2.0
4
5 import arvados_cwl
6 from arvados_cwl.arvdocker import arv_docker_clear_cache
7 import logging
8 import mock
9 import unittest
10 import os
11 import functools
12 import cwltool.process
13 from schema_salad.ref_resolver import Loader
14 from schema_salad.sourceline import cmap
15
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
23 class TestContainer(unittest.TestCase):
24
25     # The test passes no builder.resources
26     # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
27     @mock.patch("arvados.commands.keepdocker.list_images_in_arv")
28     def test_run(self, keepdocker):
29         for enable_reuse in (True, False):
30             arv_docker_clear_cache()
31
32             runner = mock.MagicMock()
33             runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
34             runner.ignore_docker_for_reuse = False
35             runner.intermediate_output_ttl = 0
36
37             keepdocker.return_value = [("zzzzz-4zz18-zzzzzzzzzzzzzz3", "")]
38             runner.api.collections().get().execute.return_value = {
39                 "portable_data_hash": "99999999999999999999999999999993+99"}
40
41             document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
42
43             tool = cmap({
44                 "inputs": [],
45                 "outputs": [],
46                 "baseCommand": "ls",
47                 "arguments": [{"valueFrom": "$(runtime.outdir)"}]
48             })
49             make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
50                                          collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
51             arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="containers", avsc_names=avsc_names,
52                                                      basedir="", make_fs_access=make_fs_access, loader=Loader({}))
53             arvtool.formatgraph = None
54             for j in arvtool.job({}, mock.MagicMock(), basedir="", name="test_run_"+str(enable_reuse),
55                                  make_fs_access=make_fs_access, tmpdir="/tmp"):
56                 j.run(enable_reuse=enable_reuse)
57                 runner.api.container_requests().create.assert_called_with(
58                     body=JsonDiffMatcher({
59                         'environment': {
60                             'HOME': '/var/spool/cwl',
61                             'TMPDIR': '/tmp'
62                         },
63                         'name': 'test_run_'+str(enable_reuse),
64                         'runtime_constraints': {
65                             'vcpus': 1,
66                             'ram': 1073741824
67                         },
68                         'use_existing': enable_reuse,
69                         'priority': 1,
70                         'mounts': {
71                             '/tmp': {'kind': 'tmp',
72                                      "capacity": 1073741824
73                                  },
74                             '/var/spool/cwl': {'kind': 'tmp',
75                                                "capacity": 1073741824 }
76                         },
77                         'state': 'Committed',
78                         'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
79                         'output_path': '/var/spool/cwl',
80                         'output_ttl': 0,
81                         'container_image': 'arvados/jobs',
82                         'command': ['ls', '/var/spool/cwl'],
83                         'cwd': '/var/spool/cwl',
84                         'scheduling_parameters': {},
85                         'properties': {},
86                     }))
87
88     # The test passes some fields in builder.resources
89     # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
90     @mock.patch("arvados.commands.keepdocker.list_images_in_arv")
91     def test_resource_requirements(self, keepdocker):
92         arv_docker_clear_cache()
93         runner = mock.MagicMock()
94         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
95         runner.ignore_docker_for_reuse = False
96         runner.intermediate_output_ttl = 3600
97         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
98
99         keepdocker.return_value = [("zzzzz-4zz18-zzzzzzzzzzzzzz3", "")]
100         runner.api.collections().get().execute.return_value = {
101             "portable_data_hash": "99999999999999999999999999999993+99"}
102
103         tool = cmap({
104             "inputs": [],
105             "outputs": [],
106             "hints": [{
107                 "class": "ResourceRequirement",
108                 "coresMin": 3,
109                 "ramMin": 3000,
110                 "tmpdirMin": 4000,
111                 "outdirMin": 5000
112             }, {
113                 "class": "http://arvados.org/cwl#RuntimeConstraints",
114                 "keep_cache": 512
115             }, {
116                 "class": "http://arvados.org/cwl#APIRequirement",
117             }, {
118                 "class": "http://arvados.org/cwl#PartitionRequirement",
119                 "partition": "blurb"
120             }, {
121                 "class": "http://arvados.org/cwl#IntermediateOutput",
122                 "outputTTL": 7200
123             }, {
124                 "class": "http://arvados.org/cwl#ReuseRequirement",
125                 "enableReuse": False
126             }],
127             "baseCommand": "ls"
128         })
129         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
130                                          collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
131         arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="containers",
132                                                  avsc_names=avsc_names, make_fs_access=make_fs_access,
133                                                  loader=Loader({}))
134         arvtool.formatgraph = None
135         for j in arvtool.job({}, mock.MagicMock(), basedir="", name="test_resource_requirements",
136                              make_fs_access=make_fs_access, tmpdir="/tmp"):
137             j.run(enable_reuse=True)
138
139         call_args, call_kwargs = runner.api.container_requests().create.call_args
140
141         call_body_expected = {
142             'environment': {
143                 'HOME': '/var/spool/cwl',
144                 'TMPDIR': '/tmp'
145             },
146             'name': 'test_resource_requirements',
147             'runtime_constraints': {
148                 'vcpus': 3,
149                 'ram': 3145728000,
150                 'keep_cache_ram': 536870912,
151                 'API': True
152             },
153             'use_existing': False,
154             'priority': 1,
155             'mounts': {
156                 '/tmp': {'kind': 'tmp',
157                          "capacity": 4194304000 },
158                 '/var/spool/cwl': {'kind': 'tmp',
159                                    "capacity": 5242880000 }
160             },
161             'state': 'Committed',
162             'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
163             'output_path': '/var/spool/cwl',
164             'output_ttl': 7200,
165             'container_image': 'arvados/jobs',
166             'command': ['ls'],
167             'cwd': '/var/spool/cwl',
168             'scheduling_parameters': {
169                 'partitions': ['blurb']
170             },
171             'properties': {}
172         }
173
174         call_body = call_kwargs.get('body', None)
175         self.assertNotEqual(None, call_body)
176         for key in call_body:
177             self.assertEqual(call_body_expected.get(key), call_body.get(key))
178
179
180     # The test passes some fields in builder.resources
181     # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
182     @mock.patch("arvados.commands.keepdocker.list_images_in_arv")
183     @mock.patch("arvados.collection.Collection")
184     def test_initial_work_dir(self, collection_mock, keepdocker):
185         arv_docker_clear_cache()
186         runner = mock.MagicMock()
187         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
188         runner.ignore_docker_for_reuse = False
189         runner.intermediate_output_ttl = 0
190         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
191
192         keepdocker.return_value = [("zzzzz-4zz18-zzzzzzzzzzzzzz3", "")]
193         runner.api.collections().get().execute.return_value = {
194             "portable_data_hash": "99999999999999999999999999999993+99"}
195
196         sourcemock = mock.MagicMock()
197         def get_collection_mock(p):
198             if "/" in p:
199                 return (sourcemock, p.split("/", 1)[1])
200             else:
201                 return (sourcemock, "")
202         runner.fs_access.get_collection.side_effect = get_collection_mock
203
204         vwdmock = mock.MagicMock()
205         collection_mock.return_value = vwdmock
206         vwdmock.portable_data_hash.return_value = "99999999999999999999999999999996+99"
207
208         tool = cmap({
209             "inputs": [],
210             "outputs": [],
211             "hints": [{
212                 "class": "InitialWorkDirRequirement",
213                 "listing": [{
214                     "class": "File",
215                     "basename": "foo",
216                     "location": "keep:99999999999999999999999999999995+99/bar"
217                 },
218                 {
219                     "class": "Directory",
220                     "basename": "foo2",
221                     "location": "keep:99999999999999999999999999999995+99"
222                 },
223                 {
224                     "class": "File",
225                     "basename": "filename",
226                     "location": "keep:99999999999999999999999999999995+99/baz/filename"
227                 },
228                 {
229                     "class": "Directory",
230                     "basename": "subdir",
231                     "location": "keep:99999999999999999999999999999995+99/subdir"
232                 }                        ]
233             }],
234             "baseCommand": "ls"
235         })
236         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
237                                          collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
238         arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="containers",
239                                                  avsc_names=avsc_names, make_fs_access=make_fs_access,
240                                                  loader=Loader({}))
241         arvtool.formatgraph = None
242         for j in arvtool.job({}, mock.MagicMock(), basedir="", name="test_initial_work_dir",
243                              make_fs_access=make_fs_access, tmpdir="/tmp"):
244             j.run()
245
246         call_args, call_kwargs = runner.api.container_requests().create.call_args
247
248         vwdmock.copy.assert_has_calls([mock.call('bar', 'foo', source_collection=sourcemock)])
249         vwdmock.copy.assert_has_calls([mock.call('', 'foo2', source_collection=sourcemock)])
250         vwdmock.copy.assert_has_calls([mock.call('baz/filename', 'filename', source_collection=sourcemock)])
251         vwdmock.copy.assert_has_calls([mock.call('subdir', 'subdir', source_collection=sourcemock)])
252
253         call_body_expected = {
254             'environment': {
255                 'HOME': '/var/spool/cwl',
256                 'TMPDIR': '/tmp'
257             },
258             'name': 'test_initial_work_dir',
259             'runtime_constraints': {
260                 'vcpus': 1,
261                 'ram': 1073741824
262             },
263             'use_existing': True,
264             'priority': 1,
265             'mounts': {
266                 '/tmp': {'kind': 'tmp',
267                          "capacity": 1073741824 },
268                 '/var/spool/cwl': {'kind': 'tmp',
269                                    "capacity": 1073741824 },
270                 '/var/spool/cwl/foo': {
271                     'kind': 'collection',
272                     'path': 'foo',
273                     'portable_data_hash': '99999999999999999999999999999996+99'
274                 },
275                 '/var/spool/cwl/foo2': {
276                     'kind': 'collection',
277                     'path': 'foo2',
278                     'portable_data_hash': '99999999999999999999999999999996+99'
279                 },
280                 '/var/spool/cwl/filename': {
281                     'kind': 'collection',
282                     'path': 'filename',
283                     'portable_data_hash': '99999999999999999999999999999996+99'
284                 },
285                 '/var/spool/cwl/subdir': {
286                     'kind': 'collection',
287                     'path': 'subdir',
288                     'portable_data_hash': '99999999999999999999999999999996+99'
289                 }
290             },
291             'state': 'Committed',
292             'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
293             'output_path': '/var/spool/cwl',
294             'output_ttl': 0,
295             'container_image': 'arvados/jobs',
296             'command': ['ls'],
297             'cwd': '/var/spool/cwl',
298             'scheduling_parameters': {
299             },
300             'properties': {}
301         }
302
303         call_body = call_kwargs.get('body', None)
304         self.assertNotEqual(None, call_body)
305         for key in call_body:
306             self.assertEqual(call_body_expected.get(key), call_body.get(key))
307
308
309     # Test redirecting stdin/stdout/stderr
310     @mock.patch("arvados.commands.keepdocker.list_images_in_arv")
311     def test_redirects(self, keepdocker):
312         arv_docker_clear_cache()
313
314         runner = mock.MagicMock()
315         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
316         runner.ignore_docker_for_reuse = False
317         runner.intermediate_output_ttl = 0
318
319         keepdocker.return_value = [("zzzzz-4zz18-zzzzzzzzzzzzzz3", "")]
320         runner.api.collections().get().execute.return_value = {
321             "portable_data_hash": "99999999999999999999999999999993+99"}
322
323         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
324
325         tool = cmap({
326             "inputs": [],
327             "outputs": [],
328             "baseCommand": "ls",
329             "stdout": "stdout.txt",
330             "stderr": "stderr.txt",
331             "stdin": "/keep/99999999999999999999999999999996+99/file.txt",
332             "arguments": [{"valueFrom": "$(runtime.outdir)"}]
333         })
334         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
335                                          collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
336         arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="containers", avsc_names=avsc_names,
337                                                  basedir="", make_fs_access=make_fs_access, loader=Loader({}))
338         arvtool.formatgraph = None
339         for j in arvtool.job({}, mock.MagicMock(), basedir="", name="test_run_redirect",
340                              make_fs_access=make_fs_access, tmpdir="/tmp"):
341             j.run()
342             runner.api.container_requests().create.assert_called_with(
343                 body=JsonDiffMatcher({
344                     'environment': {
345                         'HOME': '/var/spool/cwl',
346                         'TMPDIR': '/tmp'
347                     },
348                     'name': 'test_run_redirect',
349                     'runtime_constraints': {
350                         'vcpus': 1,
351                         'ram': 1073741824
352                     },
353                     'use_existing': True,
354                     'priority': 1,
355                     'mounts': {
356                         '/tmp': {'kind': 'tmp',
357                                  "capacity": 1073741824 },
358                         '/var/spool/cwl': {'kind': 'tmp',
359                                            "capacity": 1073741824 },
360                         "stderr": {
361                             "kind": "file",
362                             "path": "/var/spool/cwl/stderr.txt"
363                         },
364                         "stdin": {
365                             "kind": "collection",
366                             "path": "file.txt",
367                             "portable_data_hash": "99999999999999999999999999999996+99"
368                         },
369                         "stdout": {
370                             "kind": "file",
371                             "path": "/var/spool/cwl/stdout.txt"
372                         },
373                     },
374                     'state': 'Committed',
375                     'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
376                     'output_path': '/var/spool/cwl',
377                     'output_ttl': 0,
378                     'container_image': 'arvados/jobs',
379                     'command': ['ls', '/var/spool/cwl'],
380                     'cwd': '/var/spool/cwl',
381                     'scheduling_parameters': {},
382                     'properties': {},
383                 }))
384
385     @mock.patch("arvados.collection.Collection")
386     def test_done(self, col):
387         api = mock.MagicMock()
388
389         runner = mock.MagicMock()
390         runner.api = api
391         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
392         runner.num_retries = 0
393         runner.ignore_docker_for_reuse = False
394         runner.intermediate_output_ttl = 0
395
396         runner.api.containers().get().execute.return_value = {"state":"Complete",
397                                                               "output": "abc+123",
398                                                               "exit_code": 0}
399
400         col().open.return_value = []
401
402         arvjob = arvados_cwl.ArvadosContainer(runner)
403         arvjob.name = "testjob"
404         arvjob.builder = mock.MagicMock()
405         arvjob.output_callback = mock.MagicMock()
406         arvjob.collect_outputs = mock.MagicMock()
407         arvjob.successCodes = [0]
408         arvjob.outdir = "/var/spool/cwl"
409         arvjob.output_ttl = 3600
410
411         arvjob.collect_outputs.return_value = {"out": "stuff"}
412
413         arvjob.done({
414             "state": "Final",
415             "log_uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz1",
416             "output_uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2",
417             "uuid": "zzzzz-xvhdp-zzzzzzzzzzzzzzz",
418             "container_uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz",
419             "modified_at": "2017-05-26T12:01:22Z"
420         })
421
422         self.assertFalse(api.collections().create.called)
423
424         arvjob.collect_outputs.assert_called_with("keep:abc+123")
425         arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
426         runner.add_intermediate_output.assert_called_with("zzzzz-4zz18-zzzzzzzzzzzzzz2")
427
428     # The test passes no builder.resources
429     # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
430     @mock.patch("arvados.commands.keepdocker.list_images_in_arv")
431     def test_mounts(self, keepdocker):
432         arv_docker_clear_cache()
433
434         runner = mock.MagicMock()
435         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
436         runner.ignore_docker_for_reuse = False
437         runner.intermediate_output_ttl = 0
438
439         keepdocker.return_value = [("zzzzz-4zz18-zzzzzzzzzzzzzz3", "")]
440         runner.api.collections().get().execute.return_value = {
441             "portable_data_hash": "99999999999999999999999999999993+99"}
442
443         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
444
445         tool = cmap({
446             "inputs": [
447                 {"id": "p1",
448                  "type": "Directory"}
449             ],
450             "outputs": [],
451             "baseCommand": "ls",
452             "arguments": [{"valueFrom": "$(runtime.outdir)"}]
453         })
454         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
455                                      collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
456         arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="containers", avsc_names=avsc_names,
457                                                  basedir="", make_fs_access=make_fs_access, loader=Loader({}))
458         arvtool.formatgraph = None
459         job_order = {
460             "p1": {
461                 "class": "Directory",
462                 "location": "keep:99999999999999999999999999999994+44",
463                 "listing": [
464                     {
465                         "class": "File",
466                         "location": "keep:99999999999999999999999999999994+44/file1",
467                     },
468                     {
469                         "class": "File",
470                         "location": "keep:99999999999999999999999999999994+44/file2",
471                     }
472                 ]
473             }
474         }
475         for j in arvtool.job(job_order, mock.MagicMock(), basedir="", name="test_run_mounts",
476                              make_fs_access=make_fs_access, tmpdir="/tmp"):
477             j.run()
478             runner.api.container_requests().create.assert_called_with(
479                 body=JsonDiffMatcher({
480                     'environment': {
481                         'HOME': '/var/spool/cwl',
482                         'TMPDIR': '/tmp'
483                     },
484                     'name': 'test_run_mounts',
485                     'runtime_constraints': {
486                         'vcpus': 1,
487                         'ram': 1073741824
488                     },
489                     'use_existing': True,
490                     'priority': 1,
491                     'mounts': {
492                         "/keep/99999999999999999999999999999994+44": {
493                             "kind": "collection",
494                             "portable_data_hash": "99999999999999999999999999999994+44"
495                         },
496                         '/tmp': {'kind': 'tmp',
497                                  "capacity": 1073741824 },
498                         '/var/spool/cwl': {'kind': 'tmp',
499                                            "capacity": 1073741824 }
500                     },
501                     'state': 'Committed',
502                     'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
503                     'output_path': '/var/spool/cwl',
504                     'output_ttl': 0,
505                     'container_image': 'arvados/jobs',
506                     'command': ['ls', '/var/spool/cwl'],
507                     'cwd': '/var/spool/cwl',
508                     'scheduling_parameters': {},
509                     'properties': {},
510                 }))