X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/0d62edcb9d25bf4dcdb20d8872ea7b438e12fc59..0c8acc1f39413e5d9a429286dd4fd358201a48da:/sdk/cwl/tests/test_job.py diff --git a/sdk/cwl/tests/test_job.py b/sdk/cwl/tests/test_job.py index 7dbc9c8ca1..1dfd86b8c0 100644 --- a/sdk/cwl/tests/test_job.py +++ b/sdk/cwl/tests/test_job.py @@ -1,3 +1,7 @@ +# Copyright (C) The Arvados Authors. All rights reserved. +# +# SPDX-License-Identifier: Apache-2.0 + import functools import json import logging @@ -10,8 +14,11 @@ import StringIO import arvados import arvados_cwl import cwltool.process +from arvados.errors import ApiError from schema_salad.ref_resolver import Loader +from schema_salad.sourceline import cmap from .mock_discovery import get_rootDesc +from .matcher import JsonDiffMatcher, StripYAMLComments if not os.getenv('ARVADOS_DEBUG'): logging.getLogger('arvados.cwl-runner').setLevel(logging.WARN) @@ -31,22 +38,33 @@ class TestJob(unittest.TestCase): document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0") list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]] - runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"} + runner.api.collections().get().execute.return_value = {"portable_data_hash": "99999999999999999999999999999993+99"} + # Simulate reused job from another project so that we can check is a can_read + # link is added. + runner.api.jobs().create().execute.return_value = { + 'state': 'Complete' if enable_reuse else 'Queued', + 'owner_uuid': 'zzzzz-tpzed-yyyyyyyyyyyyyyy' if enable_reuse else 'zzzzz-8i9sb-zzzzzzzzzzzzzzz', + 'uuid': 'zzzzz-819sb-yyyyyyyyyyyyyyy', + 'output': None, + } - tool = { + tool = cmap({ "inputs": [], "outputs": [], "baseCommand": "ls", - "arguments": [{"valueFrom": "$(runtime.outdir)"}] - } - make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api) + "arguments": [{"valueFrom": "$(runtime.outdir)"}], + "id": "#", + "class": "CommandLineTool" + }) + 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={ + body=JsonDiffMatcher({ 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz', 'runtime_constraints': {}, 'script_parameters': { @@ -56,22 +74,40 @@ class TestJob(unittest.TestCase): }], }, 'script_version': 'master', - 'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6', + 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d', 'repository': 'arvados', 'script': 'crunchrunner', 'runtime_constraints': { - 'docker_image': 'arvados/jobs:'+arvados_cwl.__version__, + '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', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'], - ['docker_image_locator', 'in docker', 'arvados/jobs:'+arvados_cwl.__version__]] + ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'], + ['docker_image_locator', 'in docker', 'arvados/jobs']] ) + if enable_reuse: + runner.api.links().create.assert_called_with( + body=JsonDiffMatcher({ + 'link_class': 'permission', + 'name': 'can_read', + "tail_uuid": "zzzzz-8i9sb-zzzzzzzzzzzzzzz", + "head_uuid": "zzzzz-819sb-yyyyyyyyyyyyyyy", + }) + ) + # Simulate an API excepction when trying to create a + # sharing link on the job + runner.api.links().create.side_effect = ApiError( + mock.MagicMock(return_value={'status': 403}), + 'Permission denied') + j.run(enable_reuse=enable_reuse) + j.output_callback.assert_called_with({}, 'success') + else: + assert not runner.api.links().create.called # The test passes some fields in builder.resources # For the remaining fields, the defaults will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024} @@ -103,17 +139,24 @@ class TestJob(unittest.TestCase): "outputDirType": "keep_output_dir" }, { "class": "http://arvados.org/cwl#APIRequirement", + }, + { + "class": "http://arvados.org/cwl#ReuseRequirement", + "enableReuse": False }], - "baseCommand": "ls" + "baseCommand": "ls", + "id": "#", + "class": "CommandLineTool" } - 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() + j.run(enable_reuse=True) runner.api.jobs().create.assert_called_with( - body={ + body=JsonDiffMatcher({ 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz', 'runtime_constraints': {}, 'script_parameters': { @@ -124,22 +167,22 @@ class TestJob(unittest.TestCase): }] }, 'script_version': 'master', - 'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6', + 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d', 'repository': 'arvados', 'script': 'crunchrunner', 'runtime_constraints': { - 'docker_image': 'arvados/jobs:'+arvados_cwl.__version__, + '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, + }), + find_or_create=False, filters=[['repository', '=', 'arvados'], ['script', '=', 'crunchrunner'], - ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'], - ['docker_image_locator', 'in docker', 'arvados/jobs:'+arvados_cwl.__version__]]) + ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'], + ['docker_image_locator', 'in docker', 'arvados/jobs']]) @mock.patch("arvados.collection.CollectionReader") def test_done(self, reader): @@ -157,7 +200,9 @@ class TestJob(unittest.TestCase): 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"}]}) + {"items": [{"manifest_text": "XYZ"}]}, + {"items": []}, + {"items": [{"manifest_text": "ABC"}]}) arvjob = arvados_cwl.ArvadosJob(runner) arvjob.name = "testjob" @@ -175,20 +220,37 @@ class TestJob(unittest.TestCase): api.collections().list.assert_has_calls([ mock.call(), + # Output collection check mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'], ['portable_data_hash', '=', '99999999999999999999999999999993+99'], ['name', '=', 'Output 9999999 of testjob']]), mock.call().execute(num_retries=0), mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999993+99']], select=['manifest_text']), + mock.call().execute(num_retries=0), + # Log collection's turn + mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'], + ['portable_data_hash', '=', '99999999999999999999999999999994+99'], + ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]), + mock.call().execute(num_retries=0), + mock.call(limit=1, filters=[['portable_data_hash', '=', '99999999999999999999999999999994+99']], + select=['manifest_text']), mock.call().execute(num_retries=0)]) - api.collections().create.assert_called_with( - ensure_unique_name=True, - body={'portable_data_hash': '99999999999999999999999999999993+99', - 'manifest_text': 'XYZ', - 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz', - 'name': 'Output 9999999 of testjob'}) + api.collections().create.assert_has_calls([ + mock.call(ensure_unique_name=True, + body={'portable_data_hash': '99999999999999999999999999999993+99', + 'manifest_text': 'XYZ', + 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz', + 'name': 'Output 9999999 of testjob'}), + mock.call().execute(num_retries=0), + mock.call(ensure_unique_name=True, + body={'portable_data_hash': '99999999999999999999999999999994+99', + 'manifest_text': 'ABC', + 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz', + 'name': 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz'}), + mock.call().execute(num_retries=0), + ]) arvjob.output_callback.assert_called_with({"out": "stuff"}, "success") @@ -207,7 +269,10 @@ class TestJob(unittest.TestCase): 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"}]},) + api.collections().list().execute.side_effect = ( + {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]}, + {"items": [{"uuid": "zzzzz-4zz18-zzzzzzzzzzzzzz2"}]}, + ) arvjob = arvados_cwl.ArvadosJob(runner) arvjob.name = "testjob" @@ -225,10 +290,17 @@ class TestJob(unittest.TestCase): api.collections().list.assert_has_calls([ mock.call(), + # Output collection mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'], ['portable_data_hash', '=', '99999999999999999999999999999993+99'], ['name', '=', 'Output 9999999 of testjob']]), - mock.call().execute(num_retries=0)]) + mock.call().execute(num_retries=0), + # Log collection + mock.call(filters=[['owner_uuid', '=', 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'], + ['portable_data_hash', '=', '99999999999999999999999999999994+99'], + ['name', '=', 'Log of zzzzz-8i9sb-zzzzzzzzzzzzzzz']]), + mock.call().execute(num_retries=0) + ]) self.assertFalse(api.collections().create.called) @@ -238,9 +310,10 @@ class TestJob(unittest.TestCase): 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.CollectionReader") @mock.patch("arvados.collection.Collection") @mock.patch('arvados.commands.keepdocker.list_images_in_arv') - def test_run(self, list_images_in_arv, mockcollection): + def test_run(self, list_images_in_arv, mockcollection, mockcollectionreader): arvados_cwl.add_arv_hints() api = mock.MagicMock() @@ -251,18 +324,25 @@ class TestWorkflow(unittest.TestCase): list_images_in_arv.return_value = [["zzzzz-4zz18-zzzzzzzzzzzzzzz"]] runner.api.collections().get().execute.return_vaulue = {"portable_data_hash": "99999999999999999999999999999993+99"} + runner.api.collections().list().execute.return_vaulue = {"items": [{"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") + make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, + collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0)) + document_loader.fetcher_constructor = functools.partial(arvados_cwl.CollectionFetcher, api_client=api, fs_access=make_fs_access("")) + document_loader.fetcher = document_loader.fetcher_constructor(document_loader.cache, document_loader.session) + document_loader.fetch_text = document_loader.fetcher.fetch_text + document_loader.check_exists = document_loader.fetcher.check_exists + tool, metadata = document_loader.resolve_ref("tests/wf/scatter2.cwl") metadata["cwlVersion"] = tool["cwlVersion"] mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118" - make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api) 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) @@ -272,11 +352,11 @@ class TestWorkflow(unittest.TestCase): it.next().run() with open("tests/wf/scatter2_subwf.cwl") as f: - subwf = f.read() + subwf = StripYAMLComments(f.read()) 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', @@ -293,18 +373,101 @@ class TestWorkflow(unittest.TestCase): 'runtime_constraints': { 'min_scratch_mb_per_node': 2048, 'min_cores_per_node': 1, - 'docker_image': 'arvados/jobs:'+arvados_cwl.__version__, + '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'], - ['docker_image_locator', 'in docker', 'arvados/jobs:'+arvados_cwl.__version__]], + ['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}')]) + mockcollection().open().__enter__().write.assert_has_calls([mock.call( +'''{ + "fileblub": { + "basename": "token.txt", + "class": "File", + "location": "/keep/99999999999999999999999999999999+118/token.txt" + }, + "sleeptime": 5 +}''')]) + + # The test passes no builder.resources + # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024} + @mock.patch("arvados.collection.CollectionReader") + @mock.patch("arvados.collection.Collection") + @mock.patch('arvados.commands.keepdocker.list_images_in_arv') + def test_overall_resource_singlecontainer(self, list_images_in_arv, mockcollection, mockcollectionreader): + arvados_cwl.add_arv_hints() + + 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.api.collections().list().execute.return_vaulue = {"items": [{"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") + + make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, + collection_cache=arvados_cwl.CollectionCache(runner.api, None, 0)) + document_loader.fetcher_constructor = functools.partial(arvados_cwl.CollectionFetcher, api_client=api, fs_access=make_fs_access("")) + document_loader.fetcher = document_loader.fetcher_constructor(document_loader.cache, document_loader.session) + document_loader.fetch_text = document_loader.fetcher.fetch_text + document_loader.check_exists = document_loader.fetcher.check_exists + + tool, metadata = document_loader.resolve_ref("tests/wf/echo-wf.cwl") + metadata["cwlVersion"] = tool["cwlVersion"] + + mockcollection().portable_data_hash.return_value = "99999999999999999999999999999999+118" + + 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) + arvtool.formatgraph = None + it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access) + it.next().run() + it.next().run() + + with open("tests/wf/echo-subwf.cwl") as f: + subwf = StripYAMLComments(f.read()) + + runner.api.jobs().create.assert_called_with( + body=JsonDiffMatcher({ + 'minimum_script_version': 'a3f2cb186e437bfce0031b024b2157b73ed2717d', + 'repository': 'arvados', + 'script_version': 'master', + 'script': 'crunchrunner', + 'script_parameters': { + 'tasks': [{'task.env': { + 'HOME': '$(task.outdir)', + 'TMPDIR': '$(task.tmpdir)'}, + 'task.vwd': { + 'workflow.cwl': '$(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'--preserve-entire-environment', u'workflow.cwl#main', u'cwl.input.yml'], + 'task.stdout': 'cwl.output.json'}]}, + 'runtime_constraints': { + 'min_scratch_mb_per_node': 4096, + 'min_cores_per_node': 3, + 'docker_image': 'arvados/jobs', + 'min_ram_mb_per_node': 1024 + }, + 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'}), + filters=[['repository', '=', 'arvados'], + ['script', '=', 'crunchrunner'], + ['script_version', 'in git', 'a3f2cb186e437bfce0031b024b2157b73ed2717d'], + ['docker_image_locator', 'in docker', 'arvados/jobs']], + find_or_create=True) def test_default_work_api(self): arvados_cwl.add_arv_hints()