sdk/cwl/tests/test_job.py

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