X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/5c9aedbd8e623a518f5e1a92e4064a25ddd66353..ed56c26720184ca3d07f452faf014e177f1a2c2f:/sdk/cwl/tests/test_job.py diff --git a/sdk/cwl/tests/test_job.py b/sdk/cwl/tests/test_job.py index e36a6076ab..99dd3cb669 100644 --- a/sdk/cwl/tests/test_job.py +++ b/sdk/cwl/tests/test_job.py @@ -1,76 +1,96 @@ +import functools +import json import logging import mock -import unittest import os -import functools -import json +import unittest +import copy +import StringIO +import arvados import arvados_cwl import cwltool.process from schema_salad.ref_resolver import Loader +from schema_salad.sourceline import cmap +from .mock_discovery import get_rootDesc +from .matcher import JsonDiffMatcher if not os.getenv('ARVADOS_DEBUG'): logging.getLogger('arvados.cwl-runner').setLevel(logging.WARN) logging.getLogger('arvados.arv-run').setLevel(logging.WARN) - class TestJob(unittest.TestCase): # The test passes no builder.resources # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024} - def test_run(self): - runner = mock.MagicMock() - runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz" - runner.ignore_docker_for_reuse = False - document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("draft-3") + @mock.patch('arvados.commands.keepdocker.list_images_in_arv') + def test_run(self, list_images_in_arv): + for enable_reuse in (True, False): + runner = mock.MagicMock() + runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz" + runner.ignore_docker_for_reuse = False + runner.num_retries = 0 + document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0") - tool = { - "inputs": [], - "outputs": [], - "baseCommand": "ls", - "arguments": [{"valueFrom": "$(runtime.outdir)"}] - } - make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api) - arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names, - basedir="", make_fs_access=make_fs_access, loader=Loader({})) - arvtool.formatgraph = None - for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access): - j.run() - runner.api.jobs().create.assert_called_with( - body={ - 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz', - 'runtime_constraints': {}, - 'script_parameters': { - 'tasks': [{ - 'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'}, - 'command': ['ls', '$(task.outdir)'] - }], - }, - 'script_version': 'master', - 'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6', - 'repository': 'arvados', - 'script': 'crunchrunner', - 'runtime_constraints': { - 'docker_image': 'arvados/jobs', - 'min_cores_per_node': 1, - 'min_ram_mb_per_node': 1024, - 'min_scratch_mb_per_node': 2048 # tmpdirSize + outdirSize - } - }, - find_or_create=True, - filters=[['repository', '=', 'arvados'], - ['script', '=', 'crunchrunner'], - ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'], - ['docker_image_locator', 'in docker', 'arvados/jobs']] - ) + list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]] + runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"} + + tool = cmap({ + "inputs": [], + "outputs": [], + "baseCommand": "ls", + "arguments": [{"valueFrom": "$(runtime.outdir)"}] + }) + make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, + collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0)) + arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names, + basedir="", make_fs_access=make_fs_access, loader=Loader({})) + arvtool.formatgraph = None + for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access): + j.run(enable_reuse=enable_reuse) + runner.api.jobs().create.assert_called_with( + body=JsonDiffMatcher({ + 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz', + 'runtime_constraints': {}, + 'script_parameters': { + 'tasks': [{ + 'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'}, + 'command': ['ls', '$(task.outdir)'] + }], + }, + 'script_version': 'master', + 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d', + 'repository': 'arvados', + 'script': 'crunchrunner', + 'runtime_constraints': { + 'docker_image': 'arvados/jobs', + 'min_cores_per_node': 1, + 'min_ram_mb_per_node': 1024, + 'min_scratch_mb_per_node': 2048 # tmpdirSize + outdirSize + } + }), + find_or_create=enable_reuse, + filters=[['repository', '=', 'arvados'], + ['script', '=', 'crunchrunner'], + ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'], + ['docker_image_locator', 'in docker', 'arvados/jobs']] + ) # The test passes some fields in builder.resources # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024} - def test_resource_requirements(self): + @mock.patch('arvados.commands.keepdocker.list_images_in_arv') + def test_resource_requirements(self, list_images_in_arv): runner = mock.MagicMock() runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz" runner.ignore_docker_for_reuse = False - document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("draft-3") + runner.num_retries = 0 + arvados_cwl.add_arv_hints() + + list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]] + runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"} + + document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0") + tool = { "inputs": [], @@ -82,48 +102,51 @@ class TestJob(unittest.TestCase): "tmpdirMin": 4000 }, { "class": "http://arvados.org/cwl#RuntimeConstraints", - "keep_cache": 512 + "keep_cache": 512, + "outputDirType": "keep_output_dir" }, { "class": "http://arvados.org/cwl#APIRequirement", }], "baseCommand": "ls" } - make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api) + make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, + collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0)) arvtool = arvados_cwl.ArvadosCommandTool(runner, tool, work_api="jobs", avsc_names=avsc_names, make_fs_access=make_fs_access, loader=Loader({})) arvtool.formatgraph = None for j in arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access): j.run() runner.api.jobs().create.assert_called_with( - body={ + body=JsonDiffMatcher({ 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz', 'runtime_constraints': {}, 'script_parameters': { 'tasks': [{ 'task.env': {'HOME': '$(task.outdir)', 'TMPDIR': '$(task.tmpdir)'}, + 'task.keepTmpOutput': True, 'command': ['ls'] }] }, 'script_version': 'master', - 'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6', + 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d', 'repository': 'arvados', 'script': 'crunchrunner', 'runtime_constraints': { 'docker_image': 'arvados/jobs', 'min_cores_per_node': 3, - 'min_ram_mb_per_node': 3000, + 'min_ram_mb_per_node': 3512, # ramMin + keep_cache 'min_scratch_mb_per_node': 5024, # tmpdirSize + outdirSize 'keep_cache_mb_per_task': 512 } - }, + }), find_or_create=True, filters=[['repository', '=', 'arvados'], ['script', '=', 'crunchrunner'], - ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'], + ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'], ['docker_image_locator', 'in docker', 'arvados/jobs']]) - @mock.patch("arvados.collection.Collection") - def test_done(self, col): + @mock.patch("arvados.collection.CollectionReader") + def test_done(self, reader): api = mock.MagicMock() runner = mock.MagicMock() @@ -132,7 +155,11 @@ class TestJob(unittest.TestCase): runner.num_retries = 0 runner.ignore_docker_for_reuse = False - col().open.return_value = [] + reader().open.return_value = StringIO.StringIO( + """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 +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 +2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep + """) api.collections().list().execute.side_effect = ({"items": []}, {"items": [{"manifest_text": "XYZ"}]}) @@ -141,6 +168,7 @@ class TestJob(unittest.TestCase): arvjob.builder = mock.MagicMock() arvjob.output_callback = mock.MagicMock() arvjob.collect_outputs = mock.MagicMock() + arvjob.collect_outputs.return_value = {"out": "stuff"} arvjob.done({ "state": "Complete", @@ -166,8 +194,10 @@ class TestJob(unittest.TestCase): 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz', 'name': 'Output 9999999 of testjob'}) - @mock.patch("arvados.collection.Collection") - def test_done_use_existing_collection(self, col): + arvjob.output_callback.assert_called_with({"out": "stuff"}, "success") + + @mock.patch("arvados.collection.CollectionReader") + def test_done_use_existing_collection(self, reader): api = mock.MagicMock() runner = mock.MagicMock() @@ -175,7 +205,12 @@ class TestJob(unittest.TestCase): runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz" runner.num_retries = 0 - col().open.return_value = [] + reader().open.return_value = StringIO.StringIO( + """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 +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 +2016-11-02_23:12:18 c97qk-8i9sb-cryqw2blvzy4yaj 13358 0 stderr 2016/11/02 23:12:18 crunchrunner: $(task.keep)=/keep + """) + api.collections().list().execute.side_effect = ({"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]},) arvjob = arvados_cwl.ArvadosJob(runner) @@ -183,6 +218,7 @@ class TestJob(unittest.TestCase): arvjob.builder = mock.MagicMock() arvjob.output_callback = mock.MagicMock() arvjob.collect_outputs = mock.MagicMock() + arvjob.collect_outputs.return_value = {"out": "stuff"} arvjob.done({ "state": "Complete", @@ -200,17 +236,29 @@ class TestJob(unittest.TestCase): self.assertFalse(api.collections().create.called) + arvjob.output_callback.assert_called_with({"out": "stuff"}, "success") + class TestWorkflow(unittest.TestCase): # The test passes no builder.resources # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024} @mock.patch("arvados.collection.Collection") - def test_run(self, mockcollection): + @mock.patch('arvados.commands.keepdocker.list_images_in_arv') + def test_run(self, list_images_in_arv, mockcollection): arvados_cwl.add_arv_hints() - runner = arvados_cwl.ArvCwlRunner(mock.MagicMock()) + api = mock.MagicMock() + api._rootDesc = get_rootDesc() + + runner = arvados_cwl.ArvCwlRunner(api) + self.assertEqual(runner.work_api, 'jobs') + + list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]] + runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"} + runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz" runner.ignore_docker_for_reuse = False + runner.num_retries = 0 document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0") tool, metadata = document_loader.resolve_ref("tests/wf/scatter2.cwl") @@ -218,7 +266,8 @@ class TestWorkflow(unittest.TestCase): mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118" - make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api) + make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, + collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0)) arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names, basedir="", make_fs_access=make_fs_access, loader=document_loader, makeTool=runner.arv_make_tool, metadata=metadata) @@ -230,12 +279,9 @@ class TestWorkflow(unittest.TestCase): with open("tests/wf/scatter2_subwf.cwl") as f: subwf = f.read() - mockcollection().open().__enter__().write.assert_has_calls([mock.call(subwf)]) - mockcollection().open().__enter__().write.assert_has_calls([mock.call('{sleeptime: 5}')]) - runner.api.jobs().create.assert_called_with( - body={ - 'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6', + body=JsonDiffMatcher({ + 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d', 'repository': 'arvados', 'script_version': 'master', 'script': 'crunchrunner', @@ -247,7 +293,7 @@ class TestWorkflow(unittest.TestCase): 'workflow.cwl': '$(task.keep)/99999999999999999999999999999999+118/workflow.cwl', 'cwl.input.yml': '$(task.keep)/99999999999999999999999999999999+118/cwl.input.yml' }, - 'command': [u'cwltool', u'--no-container', u'--move-outputs', u'workflow.cwl#main', u'cwl.input.yml'], + 'command': [u'cwltool', u'--no-container', u'--move-outputs', u'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'], 'task.stdout': 'cwl.output.json'}]}, 'runtime_constraints': { 'min_scratch_mb_per_node': 2048, @@ -255,9 +301,24 @@ class TestWorkflow(unittest.TestCase): 'docker_image': 'arvados/jobs', 'min_ram_mb_per_node': 1024 }, - 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'}, + 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'}), filters=[['repository', '=', 'arvados'], ['script', '=', 'crunchrunner'], - ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'], + ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'], ['docker_image_locator', 'in docker', 'arvados/jobs']], find_or_create=True) + + mockcollection().open().__enter__().write.assert_has_calls([mock.call(subwf)]) + mockcollection().open().__enter__().write.assert_has_calls([mock.call( +'''{ + "sleeptime": 5 +}''')]) + + def test_default_work_api(self): + arvados_cwl.add_arv_hints() + + api = mock.MagicMock() + api._rootDesc = copy.deepcopy(get_rootDesc()) + del api._rootDesc.get('resources')['jobs']['methods']['create'] + runner = arvados_cwl.ArvCwlRunner(api) + self.assertEqual(runner.work_api, 'containers')