sdk/cwl/tests/test_job.py

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: Apache-2.0
   4
   5 import functools
   6 import json
   7 import logging
   8 import mock
   9 import os
  10 import unittest
  11 import copy
  12 import StringIO
  13
  14 import arvados
  15 import arvados_cwl
  16 import cwltool.process
  17 from arvados.errors import ApiError
  18 from schema_salad.ref_resolver import Loader
  19 from schema_salad.sourceline import cmap
  20 from .mock_discovery import get_rootDesc
  21 from .matcher import JsonDiffMatcher, StripYAMLComments
  22
  23 if not os.getenv('ARVADOS_DEBUG'):
  24     logging.getLogger('arvados.cwl-runner').setLevel(logging.WARN)
  25     logging.getLogger('arvados.arv-run').setLevel(logging.WARN)
  26
  27 class TestJob(unittest.TestCase):
  28
  29     # The test passes no builder.resources
  30     # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
  31     @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
  32     def test_run(self, list_images_in_arv):
  33         for enable_reuse in (True, False):
  34             runner = mock.MagicMock()
  35             runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
  36             runner.ignore_docker_for_reuse = False
  37             runner.num_retries = 0
  38             document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
  39
  40             list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
  41             runner.api.collections().get().execute.return_value = {"portable_data_hash": "99999999999999999999999999999993+99"}
  42             # Simulate reused job from another project so that we can check is a can_read
  43             # link is added.
  44             runner.api.jobs().create().execute.return_value = {
  45                 'state': 'Complete' if enable_reuse else 'Queued',
  46                 'owner_uuid': 'zzzzz-tpzed-yyyyyyyyyyyyyyy' if enable_reuse else 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
  47                 'uuid': 'zzzzz-819sb-yyyyyyyyyyyyyyy',
  48                 'output': None,
  49             }
  50
  51             tool = cmap({
  52                 "inputs": [],
  53                 "outputs": [],
  54                 "baseCommand": "ls",
  55                 "arguments": [{"valueFrom": "$(runtime.outdir)"}]
  56             })
  57             make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
  58                                          collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
  59             arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
  60                                                      basedir="", make_fs_access=make_fs_access, loader=Loader({}))
  61             arvtool.formatgraph = None
  62             for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
  63                 j.run(enable_reuse=enable_reuse)
  64                 runner.api.jobs().create.assert_called_with(
  65                     body=JsonDiffMatcher({
  66                         'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
  67                         'runtime_constraints': {},
  68                         'script_parameters': {
  69                             'tasks': [{
  70                                 'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
  71                                 'command': ['ls', '$(task.outdir)']
  72                             }],
  73                         },
  74                         'script_version': 'master',
  75                         'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
  76                         'repository': 'arvados',
  77                         'script': 'crunchrunner',
  78                         'runtime_constraints': {
  79                             'docker_image': 'arvados/jobs',
  80                             'min_cores_per_node': 1,
  81                             'min_ram_mb_per_node': 1024,
  82                             'min_scratch_mb_per_node': 2048 # tmpdirSize + outdirSize
  83                         }
  84                     }),
  85                     find_or_create=enable_reuse,
  86                     filters=[['repository', '=', 'arvados'],
  87                              ['script', '=', 'crunchrunner'],
  88                              ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
  89                              ['docker_image_locator', 'in docker', 'arvados/jobs']]
  90                 )
  91                 if enable_reuse:
  92                     runner.api.links().create.assert_called_with(
  93                         body=JsonDiffMatcher({
  94                             'link_class': 'permission',
  95                             'name': 'can_read',
  96                             "tail_uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz",
  97                             "head_uuid": "zzzzz-819sb-yyyyyyyyyyyyyyy",
  98                         })
  99                     )
 100                     # Simulate an API excepction when trying to create a
 101                     # sharing link on the job
 102                     runner.api.links().create.side_effect = ApiError(
 103                         mock.MagicMock(return_value={'status': 403}),
 104                         'Permission denied')
 105                     j.run(enable_reuse=enable_reuse)
 106                     j.output_callback.assert_called_with({}, 'success')
 107                 else:
 108                     assert not runner.api.links().create.called
 109
 110     # The test passes some fields in builder.resources
 111     # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
 112     @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
 113     def test_resource_requirements(self, list_images_in_arv):
 114         runner = mock.MagicMock()
 115         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
 116         runner.ignore_docker_for_reuse = False
 117         runner.num_retries = 0
 118         arvados_cwl.add_arv_hints()
 119
 120         list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
 121         runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
 122
 123         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
 124
 125
 126         tool = {
 127             "inputs": [],
 128             "outputs": [],
 129             "hints": [{
 130                 "class": "ResourceRequirement",
 131                 "coresMin": 3,
 132                 "ramMin": 3000,
 133                 "tmpdirMin": 4000
 134             }, {
 135                 "class": "http://arvados.org/cwl#RuntimeConstraints",
 136                 "keep_cache": 512,
 137                 "outputDirType": "keep_output_dir"
 138             }, {
 139                 "class": "http://arvados.org/cwl#APIRequirement",
 140             },
 141             {
 142                 "class": "http://arvados.org/cwl#ReuseRequirement",
 143                 "enableReuse": False
 144             }],
 145             "baseCommand": "ls"
 146         }
 147         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
 148                                          collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
 149         arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
 150                                                  make_fs_access=make_fs_access, loader=Loader({}))
 151         arvtool.formatgraph = None
 152         for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
 153             j.run(enable_reuse=True)
 154         runner.api.jobs().create.assert_called_with(
 155             body=JsonDiffMatcher({
 156                 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
 157                 'runtime_constraints': {},
 158                 'script_parameters': {
 159                     'tasks': [{
 160                         'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
 161                         'task.keepTmpOutput': True,
 162                         'command': ['ls']
 163                     }]
 164             },
 165             'script_version': 'master',
 166                 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
 167                 'repository': 'arvados',
 168                 'script': 'crunchrunner',
 169                 'runtime_constraints': {
 170                     'docker_image': 'arvados/jobs',
 171                     'min_cores_per_node': 3,
 172                     'min_ram_mb_per_node': 3512,     # ramMin + keep_cache
 173                     'min_scratch_mb_per_node': 5024, # tmpdirSize + outdirSize
 174                     'keep_cache_mb_per_task': 512
 175                 }
 176             }),
 177             find_or_create=False,
 178             filters=[['repository', '=', 'arvados'],
 179                      ['script', '=', 'crunchrunner'],
 180                      ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
 181                      ['docker_image_locator', 'in docker', 'arvados/jobs']])
 182
 183     @mock.patch("arvados.collection.CollectionReader")
 184     def test_done(self, reader):
 185         api = mock.MagicMock()
 186
 187         runner = mock.MagicMock()
 188         runner.api = api
 189         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
 190         runner.num_retries = 0
 191         runner.ignore_docker_for_reuse = False
 192
 193         reader().open.return_value = StringIO.StringIO(
 194             """2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.tmpdir)=/tmp/crunch-job-task-work/compute3.1/tmpdir
 195 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.outdir)=/tmp/crunch-job-task-work/compute3.1/outdir
 196 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
 197         """)
 198         api.collections().list().execute.side_effect = ({"items": []},
 199                                                         {"items": [{"manifest_text": "XYZ"}]},
 200                                                         {"items": []},
 201                                                         {"items": [{"manifest_text": "ABC"}]})
 202
 203         arvjob = arvados_cwl.ArvadosJob(runner)
 204         arvjob.name = "testjob"
 205         arvjob.builder = mock.MagicMock()
 206         arvjob.output_callback = mock.MagicMock()
 207         arvjob.collect_outputs = mock.MagicMock()
 208         arvjob.collect_outputs.return_value = {"out": "stuff"}
 209
 210         arvjob.done({
 211             "state": "Complete",
 212             "output": "99999999999999999999999999999993+99",
 213             "log": "99999999999999999999999999999994+99",
 214             "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
 215         })
 216
 217         api.collections().list.assert_has_calls([
 218             mock.call(),
 219             # Output collection check
 220             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
 221                           ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
 222                           ['name', '=', 'Output 9999999 of testjob']]),
 223             mock.call().execute(num_retries=0),
 224             mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999993+99']],
 225                  select=['manifest_text']),
 226             mock.call().execute(num_retries=0),
 227             # Log collection's turn
 228             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
 229                           ['portable_data_hash', '=', '99999999999999999999999999999994+99'],
 230                           ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]),
 231             mock.call().execute(num_retries=0),
 232             mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999994+99']],
 233                  select=['manifest_text']),
 234             mock.call().execute(num_retries=0)])
 235
 236         api.collections().create.assert_has_calls([
 237             mock.call(ensure_unique_name=True,
 238                       body={'portable_data_hash': '99999999999999999999999999999993+99',
 239                             'manifest_text': 'XYZ',
 240                             'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
 241                             'name': 'Output 9999999 of testjob'}),
 242             mock.call().execute(num_retries=0),
 243             mock.call(ensure_unique_name=True,
 244                       body={'portable_data_hash': '99999999999999999999999999999994+99',
 245                             'manifest_text': 'ABC',
 246                             'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
 247                             'name': 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
 248             mock.call().execute(num_retries=0),
 249         ])
 250
 251         arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
 252
 253     @mock.patch("arvados.collection.CollectionReader")
 254     def test_done_use_existing_collection(self, reader):
 255         api = mock.MagicMock()
 256
 257         runner = mock.MagicMock()
 258         runner.api = api
 259         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
 260         runner.num_retries = 0
 261
 262         reader().open.return_value = StringIO.StringIO(
 263             """2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.tmpdir)=/tmp/crunch-job-task-work/compute3.1/tmpdir
 264 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.outdir)=/tmp/crunch-job-task-work/compute3.1/outdir
 265 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
 266         """)
 267
 268         api.collections().list().execute.side_effect = (
 269             {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},
 270             {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},
 271         )
 272
 273         arvjob = arvados_cwl.ArvadosJob(runner)
 274         arvjob.name = "testjob"
 275         arvjob.builder = mock.MagicMock()
 276         arvjob.output_callback = mock.MagicMock()
 277         arvjob.collect_outputs = mock.MagicMock()
 278         arvjob.collect_outputs.return_value = {"out": "stuff"}
 279
 280         arvjob.done({
 281             "state": "Complete",
 282             "output": "99999999999999999999999999999993+99",
 283             "log": "99999999999999999999999999999994+99",
 284             "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
 285         })
 286
 287         api.collections().list.assert_has_calls([
 288             mock.call(),
 289             # Output collection
 290             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
 291                                ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
 292                                ['name', '=', 'Output 9999999 of testjob']]),
 293             mock.call().execute(num_retries=0),
 294             # Log collection
 295             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
 296                                ['portable_data_hash', '=', '99999999999999999999999999999994+99'],
 297                                ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]),
 298             mock.call().execute(num_retries=0)
 299         ])
 300
 301         self.assertFalse(api.collections().create.called)
 302
 303         arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
 304
 305
 306 class TestWorkflow(unittest.TestCase):
 307     # The test passes no builder.resources
 308     # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
 309     @mock.patch("arvados.collection.CollectionReader")
 310     @mock.patch("arvados.collection.Collection")
 311     @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
 312     def test_run(self, list_images_in_arv, mockcollection, mockcollectionreader):
 313         arvados_cwl.add_arv_hints()
 314
 315         api = mock.MagicMock()
 316         api._rootDesc = get_rootDesc()
 317
 318         runner = arvados_cwl.ArvCwlRunner(api)
 319         self.assertEqual(runner.work_api, 'jobs')
 320
 321         list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
 322         runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
 323         runner.api.collections().list().execute.return_vaulue = {"items": [{"portable_data_hash": "99999999999999999999999999999993+99"}]}
 324
 325         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
 326         runner.ignore_docker_for_reuse = False
 327         runner.num_retries = 0
 328         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
 329
 330         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess,
 331                                          collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0))
 332         document_loader.fetcher_constructor = functools.partial(arvados_cwl.CollectionFetcher, api_client=api, fs_access=make_fs_access(""))
 333         document_loader.fetcher = document_loader.fetcher_constructor(document_loader.cache, document_loader.session)
 334         document_loader.fetch_text = document_loader.fetcher.fetch_text
 335         document_loader.check_exists = document_loader.fetcher.check_exists
 336
 337         tool, metadata = document_loader.resolve_ref("tests/wf/scatter2.cwl")
 338         metadata["cwlVersion"] = tool["cwlVersion"]
 339
 340         mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118"
 341
 342         arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names,
 343                                               basedir="", make_fs_access=make_fs_access, loader=document_loader,
 344                                               makeTool=runner.arv_make_tool, metadata=metadata)
 345         arvtool.formatgraph = None
 346         it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access)
 347         it.next().run()
 348         it.next().run()
 349
 350         with open("tests/wf/scatter2_subwf.cwl") as f:
 351             subwf = StripYAMLComments(f.read())
 352
 353         runner.api.jobs().create.assert_called_with(
 354             body=JsonDiffMatcher({
 355                 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d',
 356                 'repository': 'arvados',
 357                 'script_version': 'master',
 358                 'script': 'crunchrunner',
 359                 'script_parameters': {
 360                     'tasks': [{'task.env': {
 361                         'HOME': '$(task.outdir)',
 362                         'TMPDIR': '$(task.tmpdir)'},
 363                                'task.vwd': {
 364                                    'workflow.cwl': '$(task.keep)/99999999999999999999999999999999+118/workflow.cwl',
 365                                    'cwl.input.yml': '$(task.keep)/99999999999999999999999999999999+118/cwl.input.yml'
 366                                },
 367                     'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'],
 368                     'task.stdout': 'cwl.output.json'}]},
 369                 'runtime_constraints': {
 370                     'min_scratch_mb_per_node': 2048,
 371                     'min_cores_per_node': 1,
 372                     'docker_image': 'arvados/jobs',
 373                     'min_ram_mb_per_node': 1024
 374                 },
 375                 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'}),
 376             filters=[['repository', '=', 'arvados'],
 377                      ['script', '=', 'crunchrunner'],
 378                      ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'],
 379                      ['docker_image_locator', 'in docker', 'arvados/jobs']],
 380             find_or_create=True)
 381
 382         mockcollection().open().__enter__().write.assert_has_calls([mock.call(subwf)])
 383         mockcollection().open().__enter__().write.assert_has_calls([mock.call(
 384 '''{
 385   "fileblub": {
 386     "basename": "token.txt",
 387     "class": "File",
 388     "location": "/keep/99999999999999999999999999999999+118/token.txt"
 389   },
 390   "sleeptime": 5
 391 }''')])
 392
 393     def test_default_work_api(self):
 394         arvados_cwl.add_arv_hints()
 395
 396         api = mock.MagicMock()
 397         api._rootDesc = copy.deepcopy(get_rootDesc())
 398         del api._rootDesc.get('resources')['jobs']['methods']['create']
 399         runner = arvados_cwl.ArvCwlRunner(api)
 400         self.assertEqual(runner.work_api, 'containers')