10032: Fix test to include --preserve-entire-environment
[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         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("draft-3")
26
27         tool = {
28             "inputs": [],
29             "outputs": [],
30             "baseCommand": "ls",
31             "arguments": [{"valueFrom": "$(runtime.outdir)"}]
32         }
33         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
34         arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
35                                                  basedir="", make_fs_access=make_fs_access, loader=Loader({}))
36         arvtool.formatgraph = None
37         for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
38             j.run()
39             runner.api.jobs().create.assert_called_with(
40                 body={
41                     'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
42                     'runtime_constraints': {},
43                     'script_parameters': {
44                         'tasks': [{
45                             'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
46                             'command': ['ls', '$(task.outdir)']
47                         }],
48                     },
49                     'script_version': 'master',
50                     'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6',
51                     'repository': 'arvados',
52                     'script': 'crunchrunner',
53                     'runtime_constraints': {
54                         'docker_image': 'arvados/jobs',
55                         'min_cores_per_node': 1,
56                         'min_ram_mb_per_node': 1024,
57                         'min_scratch_mb_per_node': 2048 # tmpdirSize + outdirSize
58                     }
59                 },
60                 find_or_create=True,
61                 filters=[['repository', '=', 'arvados'],
62                          ['script', '=', 'crunchrunner'],
63                          ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'],
64                          ['docker_image_locator', 'in docker', 'arvados/jobs']]
65             )
66
67     # The test passes some fields in builder.resources
68     # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
69     def test_resource_requirements(self):
70         runner = mock.MagicMock()
71         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
72         runner.ignore_docker_for_reuse = False
73         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("draft-3")
74
75         tool = {
76             "inputs": [],
77             "outputs": [],
78             "hints": [{
79                 "class": "ResourceRequirement",
80                 "coresMin": 3,
81                 "ramMin": 3000,
82                 "tmpdirMin": 4000
83             }, {
84                 "class": "http://arvados.org/cwl#RuntimeConstraints",
85                 "keep_cache": 512
86             }, {
87                 "class": "http://arvados.org/cwl#APIRequirement",
88             }],
89             "baseCommand": "ls"
90         }
91         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
92         arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
93                                                  make_fs_access=make_fs_access, loader=Loader({}))
94         arvtool.formatgraph = None
95         for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
96             j.run()
97         runner.api.jobs().create.assert_called_with(
98             body={
99                 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
100                 'runtime_constraints': {},
101                 'script_parameters': {
102                     'tasks': [{
103                         'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
104                         'command': ['ls']
105                     }]
106             },
107             'script_version': 'master',
108                 'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6',
109                 'repository': 'arvados',
110                 'script': 'crunchrunner',
111                 'runtime_constraints': {
112                     'docker_image': 'arvados/jobs',
113                     'min_cores_per_node': 3,
114                     'min_ram_mb_per_node': 3000,
115                     'min_scratch_mb_per_node': 5024, # tmpdirSize + outdirSize
116                     'keep_cache_mb_per_task': 512
117                 }
118             },
119             find_or_create=True,
120             filters=[['repository', '=', 'arvados'],
121                      ['script', '=', 'crunchrunner'],
122                      ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'],
123                      ['docker_image_locator', 'in docker', 'arvados/jobs']])
124
125     @mock.patch("arvados.collection.Collection")
126     def test_done(self, col):
127         api = mock.MagicMock()
128
129         runner = mock.MagicMock()
130         runner.api = api
131         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
132         runner.num_retries = 0
133         runner.ignore_docker_for_reuse = False
134
135         col().open.return_value = []
136         api.collections().list().execute.side_effect = ({"items": []},
137                                                         {"items": [{"manifest_text": "XYZ"}]})
138
139         arvjob = arvados_cwl.ArvadosJob(runner)
140         arvjob.name = "testjob"
141         arvjob.builder = mock.MagicMock()
142         arvjob.output_callback = mock.MagicMock()
143         arvjob.collect_outputs = mock.MagicMock()
144
145         arvjob.done({
146             "state": "Complete",
147             "output": "99999999999999999999999999999993+99",
148             "log": "99999999999999999999999999999994+99",
149             "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
150         })
151
152         api.collections().list.assert_has_calls([
153             mock.call(),
154             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
155                           ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
156                           ['name', '=', 'Output 9999999 of testjob']]),
157             mock.call().execute(num_retries=0),
158             mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999993+99']],
159                  select=['manifest_text']),
160             mock.call().execute(num_retries=0)])
161
162         api.collections().create.assert_called_with(
163             ensure_unique_name=True,
164             body={'portable_data_hash': '99999999999999999999999999999993+99',
165                   'manifest_text': 'XYZ',
166                   'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
167                   'name': 'Output 9999999 of testjob'})
168
169     @mock.patch("arvados.collection.Collection")
170     def test_done_use_existing_collection(self, col):
171         api = mock.MagicMock()
172
173         runner = mock.MagicMock()
174         runner.api = api
175         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
176         runner.num_retries = 0
177
178         col().open.return_value = []
179         api.collections().list().execute.side_effect = ({"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},)
180
181         arvjob = arvados_cwl.ArvadosJob(runner)
182         arvjob.name = "testjob"
183         arvjob.builder = mock.MagicMock()
184         arvjob.output_callback = mock.MagicMock()
185         arvjob.collect_outputs = mock.MagicMock()
186
187         arvjob.done({
188             "state": "Complete",
189             "output": "99999999999999999999999999999993+99",
190             "log": "99999999999999999999999999999994+99",
191             "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
192         })
193
194         api.collections().list.assert_has_calls([
195             mock.call(),
196             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
197                                ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
198                                ['name', '=', 'Output 9999999 of testjob']]),
199             mock.call().execute(num_retries=0)])
200
201         self.assertFalse(api.collections().create.called)
202
203
204 class TestWorkflow(unittest.TestCase):
205     # The test passes no builder.resources
206     # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
207     @mock.patch("arvados.collection.Collection")
208     def test_run(self, mockcollection):
209         arvados_cwl.add_arv_hints()
210
211         runner = arvados_cwl.ArvCwlRunner(mock.MagicMock())
212         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
213         runner.ignore_docker_for_reuse = False
214         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
215
216         tool, metadata = document_loader.resolve_ref("tests/wf/scatter2.cwl")
217         metadata["cwlVersion"] = tool["cwlVersion"]
218
219         mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118"
220
221         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
222         arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names,
223                                               basedir="", make_fs_access=make_fs_access, loader=document_loader,
224                                               makeTool=runner.arv_make_tool, metadata=metadata)
225         arvtool.formatgraph = None
226         it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access)
227         it.next().run()
228         it.next().run()
229
230         with open("tests/wf/scatter2_subwf.cwl") as f:
231             subwf = f.read()
232
233         mockcollection().open().__enter__().write.assert_has_calls([mock.call(subwf)])
234         mockcollection().open().__enter__().write.assert_has_calls([mock.call('{sleeptime: 5}')])
235
236         runner.api.jobs().create.assert_called_with(
237             body={
238                 'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6',
239                 'repository': 'arvados',
240                 'script_version': 'master',
241                 'script': 'crunchrunner',
242                 'script_parameters': {
243                     'tasks': [{'task.env': {
244                         'HOME': '$(task.outdir)',
245                         'TMPDIR': '$(task.tmpdir)'},
246                                'task.vwd': {
247                                    'workflow.cwl': '$(task.keep)/99999999999999999999999999999999+118/workflow.cwl',
248                                    'cwl.input.yml': '$(task.keep)/99999999999999999999999999999999+118/cwl.input.yml'
249                                },
250                     'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'],
251                     'task.stdout': 'cwl.output.json'}]},
252                 'runtime_constraints': {
253                     'min_scratch_mb_per_node': 2048,
254                     'min_cores_per_node': 1,
255                     'docker_image': 'arvados/jobs',
256                     'min_ram_mb_per_node': 1024
257                 },
258                 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'},
259             filters=[['repository', '=', 'arvados'],
260                      ['script', '=', 'crunchrunner'],
261                      ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'],
262                      ['docker_image_locator', 'in docker', 'arvados/jobs']],
263             find_or_create=True)