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 schema_salad.ref_resolver import Loader
  14 from .mock_discovery import get_rootDesc
  15
  16 if not os.getenv('ARVADOS_DEBUG'):
  17     logging.getLogger('arvados.cwl-runner').setLevel(logging.WARN)
  18     logging.getLogger('arvados.arv-run').setLevel(logging.WARN)
  19
  20 class TestJob(unittest.TestCase):
  21
  22     # The test passes no builder.resources
  23     # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
  24     @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
  25     def test_run(self, list_images_in_arv):
  26         for enable_reuse in (True, False):
  27             runner = mock.MagicMock()
  28             runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
  29             runner.ignore_docker_for_reuse = False
  30             runner.num_retries = 0
  31             document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
  32
  33             list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
  34             runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
  35
  36             tool = {
  37                 "inputs": [],
  38                 "outputs": [],
  39                 "baseCommand": "ls",
  40                 "arguments": [{"valueFrom": "$(runtime.outdir)"}]
  41             }
  42             make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
  43             arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
  44                                                      basedir="", make_fs_access=make_fs_access, loader=Loader({}))
  45             arvtool.formatgraph = None
  46             for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
  47                 j.run(enable_reuse=enable_reuse)
  48                 runner.api.jobs().create.assert_called_with(
  49                     body={
  50                         'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
  51                         'runtime_constraints': {},
  52                         'script_parameters': {
  53                             'tasks': [{
  54                                 'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
  55                                 'command': ['ls', '$(task.outdir)']
  56                             }],
  57                         },
  58                         'script_version': 'master',
  59                         'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6',
  60                         'repository': 'arvados',
  61                         'script': 'crunchrunner',
  62                         'runtime_constraints': {
  63                             'docker_image': 'arvados/jobs:'+arvados_cwl.__version__,
  64                             'min_cores_per_node': 1,
  65                             'min_ram_mb_per_node': 1024,
  66                             'min_scratch_mb_per_node': 2048 # tmpdirSize + outdirSize
  67                         }
  68                     },
  69                     find_or_create=enable_reuse,
  70                     filters=[['repository', '=', 'arvados'],
  71                              ['script', '=', 'crunchrunner'],
  72                              ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'],
  73                              ['docker_image_locator', 'in docker', 'arvados/jobs:'+arvados_cwl.__version__]]
  74                 )
  75
  76     # The test passes some fields in builder.resources
  77     # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
  78     @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
  79     def test_resource_requirements(self, list_images_in_arv):
  80         runner = mock.MagicMock()
  81         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
  82         runner.ignore_docker_for_reuse = False
  83         runner.num_retries = 0
  84         arvados_cwl.add_arv_hints()
  85
  86         list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
  87         runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
  88
  89         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
  90
  91
  92         tool = {
  93             "inputs": [],
  94             "outputs": [],
  95             "hints": [{
  96                 "class": "ResourceRequirement",
  97                 "coresMin": 3,
  98                 "ramMin": 3000,
  99                 "tmpdirMin": 4000
 100             }, {
 101                 "class": "http://arvados.org/cwl#RuntimeConstraints",
 102                 "keep_cache": 512,
 103                 "outputDirType": "keep_output_dir"
 104             }, {
 105                 "class": "http://arvados.org/cwl#APIRequirement",
 106             }],
 107             "baseCommand": "ls"
 108         }
 109         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
 110         arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names,
 111                                                  make_fs_access=make_fs_access, loader=Loader({}))
 112         arvtool.formatgraph = None
 113         for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access):
 114             j.run()
 115         runner.api.jobs().create.assert_called_with(
 116             body={
 117                 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
 118                 'runtime_constraints': {},
 119                 'script_parameters': {
 120                     'tasks': [{
 121                         'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'},
 122                         'task.keepTmpOutput': True,
 123                         'command': ['ls']
 124                     }]
 125             },
 126             'script_version': 'master',
 127                 'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6',
 128                 'repository': 'arvados',
 129                 'script': 'crunchrunner',
 130                 'runtime_constraints': {
 131                     'docker_image': 'arvados/jobs:'+arvados_cwl.__version__,
 132                     'min_cores_per_node': 3,
 133                     'min_ram_mb_per_node': 3000,
 134                     'min_scratch_mb_per_node': 5024, # tmpdirSize + outdirSize
 135                     'keep_cache_mb_per_task': 512
 136                 }
 137             },
 138             find_or_create=True,
 139             filters=[['repository', '=', 'arvados'],
 140                      ['script', '=', 'crunchrunner'],
 141                      ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'],
 142                      ['docker_image_locator', 'in docker', 'arvados/jobs:'+arvados_cwl.__version__]])
 143
 144     @mock.patch("arvados.collection.CollectionReader")
 145     def test_done(self, reader):
 146         api = mock.MagicMock()
 147
 148         runner = mock.MagicMock()
 149         runner.api = api
 150         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
 151         runner.num_retries = 0
 152         runner.ignore_docker_for_reuse = False
 153
 154         reader().open.return_value = StringIO.StringIO(
 155             """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
 156 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
 157 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
 158         """)
 159         api.collections().list().execute.side_effect = ({"items": []},
 160                                                         {"items": [{"manifest_text": "XYZ"}]})
 161
 162         arvjob = arvados_cwl.ArvadosJob(runner)
 163         arvjob.name = "testjob"
 164         arvjob.builder = mock.MagicMock()
 165         arvjob.output_callback = mock.MagicMock()
 166         arvjob.collect_outputs = mock.MagicMock()
 167         arvjob.collect_outputs.return_value = {"out": "stuff"}
 168
 169         arvjob.done({
 170             "state": "Complete",
 171             "output": "99999999999999999999999999999993+99",
 172             "log": "99999999999999999999999999999994+99",
 173             "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
 174         })
 175
 176         api.collections().list.assert_has_calls([
 177             mock.call(),
 178             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
 179                           ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
 180                           ['name', '=', 'Output 9999999 of testjob']]),
 181             mock.call().execute(num_retries=0),
 182             mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999993+99']],
 183                  select=['manifest_text']),
 184             mock.call().execute(num_retries=0)])
 185
 186         api.collections().create.assert_called_with(
 187             ensure_unique_name=True,
 188             body={'portable_data_hash': '99999999999999999999999999999993+99',
 189                   'manifest_text': 'XYZ',
 190                   'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz',
 191                   'name': 'Output 9999999 of testjob'})
 192
 193         arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
 194
 195     @mock.patch("arvados.collection.CollectionReader")
 196     def test_done_use_existing_collection(self, reader):
 197         api = mock.MagicMock()
 198
 199         runner = mock.MagicMock()
 200         runner.api = api
 201         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
 202         runner.num_retries = 0
 203
 204         reader().open.return_value = StringIO.StringIO(
 205             """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
 206 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
 207 2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep
 208         """)
 209
 210         api.collections().list().execute.side_effect = ({"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},)
 211
 212         arvjob = arvados_cwl.ArvadosJob(runner)
 213         arvjob.name = "testjob"
 214         arvjob.builder = mock.MagicMock()
 215         arvjob.output_callback = mock.MagicMock()
 216         arvjob.collect_outputs = mock.MagicMock()
 217         arvjob.collect_outputs.return_value = {"out": "stuff"}
 218
 219         arvjob.done({
 220             "state": "Complete",
 221             "output": "99999999999999999999999999999993+99",
 222             "log": "99999999999999999999999999999994+99",
 223             "uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
 224         })
 225
 226         api.collections().list.assert_has_calls([
 227             mock.call(),
 228             mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'],
 229                                ['portable_data_hash', '=', '99999999999999999999999999999993+99'],
 230                                ['name', '=', 'Output 9999999 of testjob']]),
 231             mock.call().execute(num_retries=0)])
 232
 233         self.assertFalse(api.collections().create.called)
 234
 235         arvjob.output_callback.assert_called_with({"out": "stuff"}, "success")
 236
 237
 238 class TestWorkflow(unittest.TestCase):
 239     # The test passes no builder.resources
 240     # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
 241     @mock.patch("arvados.collection.Collection")
 242     @mock.patch('arvados.commands.keepdocker.list_images_in_arv')
 243     def test_run(self, list_images_in_arv, mockcollection):
 244         arvados_cwl.add_arv_hints()
 245
 246         api = mock.MagicMock()
 247         api._rootDesc = get_rootDesc()
 248
 249         runner = arvados_cwl.ArvCwlRunner(api)
 250         self.assertEqual(runner.work_api, 'jobs')
 251
 252         list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]]
 253         runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"}
 254
 255         runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
 256         runner.ignore_docker_for_reuse = False
 257         runner.num_retries = 0
 258         document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
 259
 260         tool, metadata = document_loader.resolve_ref("tests/wf/scatter2.cwl")
 261         metadata["cwlVersion"] = tool["cwlVersion"]
 262
 263         mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118"
 264
 265         make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
 266         arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names,
 267                                               basedir="", make_fs_access=make_fs_access, loader=document_loader,
 268                                               makeTool=runner.arv_make_tool, metadata=metadata)
 269         arvtool.formatgraph = None
 270         it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access)
 271         it.next().run()
 272         it.next().run()
 273
 274         with open("tests/wf/scatter2_subwf.cwl") as f:
 275             subwf = f.read()
 276
 277         runner.api.jobs().create.assert_called_with(
 278             body={
 279                 'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6',
 280                 'repository': 'arvados',
 281                 'script_version': 'master',
 282                 'script': 'crunchrunner',
 283                 'script_parameters': {
 284                     'tasks': [{'task.env': {
 285                         'HOME': '$(task.outdir)',
 286                         'TMPDIR': '$(task.tmpdir)'},
 287                                'task.vwd': {
 288                                    'workflow.cwl': '$(task.keep)/99999999999999999999999999999999+118/workflow.cwl',
 289                                    'cwl.input.yml': '$(task.keep)/99999999999999999999999999999999+118/cwl.input.yml'
 290                                },
 291                     'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'],
 292                     'task.stdout': 'cwl.output.json'}]},
 293                 'runtime_constraints': {
 294                     'min_scratch_mb_per_node': 2048,
 295                     'min_cores_per_node': 1,
 296                     'docker_image': 'arvados/jobs:'+arvados_cwl.__version__,
 297                     'min_ram_mb_per_node': 1024
 298                 },
 299                 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'},
 300             filters=[['repository', '=', 'arvados'],
 301                      ['script', '=', 'crunchrunner'],
 302                      ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'],
 303                      ['docker_image_locator', 'in docker', 'arvados/jobs:'+arvados_cwl.__version__]],
 304             find_or_create=True)
 305
 306         mockcollection().open().__enter__().write.assert_has_calls([mock.call(subwf)])
 307         mockcollection().open().__enter__().write.assert_has_calls([mock.call('{sleeptime: 5}')])
 308
 309     def test_default_work_api(self):
 310         arvados_cwl.add_arv_hints()
 311
 312         api = mock.MagicMock()
 313         api._rootDesc = copy.deepcopy(get_rootDesc())
 314         del api._rootDesc.get('resources')['jobs']['methods']['create']
 315         runner = arvados_cwl.ArvCwlRunner(api)
 316         self.assertEqual(runner.work_api, 'containers')