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