X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/c7399ec7afdf0cfdd0f3177f410f102083a26e15..04eca602e1bda34cfcfdd518d322050723de10d0:/sdk/cwl/arvados_cwl/__init__.py diff --git a/sdk/cwl/arvados_cwl/__init__.py b/sdk/cwl/arvados_cwl/__init__.py index d15acf767f..5c5f5502d0 100644 --- a/sdk/cwl/arvados_cwl/__init__.py +++ b/sdk/cwl/arvados_cwl/__init__.py @@ -22,8 +22,8 @@ from cwltool.errors import WorkflowException import cwltool.main import cwltool.workflow import cwltool.process -import schema_salad from schema_salad.sourceline import SourceLine +import schema_salad.validate as validate import arvados import arvados.config @@ -43,7 +43,7 @@ from ._version import __version__ from cwltool.pack import pack from cwltool.process import shortname, UnsupportedRequirement, use_custom_schema from cwltool.pathmapper import adjustFileObjs, adjustDirObjs, get_listing -from cwltool.draft2tool import compute_checksums +from cwltool.command_line_tool import compute_checksums from arvados.api import OrderedJsonModel logger = logging.getLogger('arvados.cwl-runner') @@ -54,6 +54,8 @@ arvados.log_handler.setFormatter(logging.Formatter( '%(asctime)s %(name)s %(levelname)s: %(message)s', '%Y-%m-%d %H:%M:%S')) +DEFAULT_PRIORITY = 500 + class ArvCwlRunner(object): """Execute a CWL tool or workflow, submit work (using either jobs or containers API), wait for them to complete, and report output. @@ -111,8 +113,7 @@ class ArvCwlRunner(object): kwargs["fetcher_constructor"] = partial(CollectionFetcher, api_client=self.api, fs_access=CollectionFsAccess("", collection_cache=self.collection_cache), - num_retries=self.num_retries, - overrides=kwargs.get("override_tools")) + num_retries=self.num_retries) kwargs["resolver"] = partial(collectionResolver, self.api, num_retries=self.num_retries) if "class" in toolpath_object and toolpath_object["class"] == "CommandLineTool": return ArvadosCommandTool(self, toolpath_object, **kwargs) @@ -226,13 +227,16 @@ class ArvCwlRunner(object): def check_features(self, obj): if isinstance(obj, dict): - if obj.get("writable"): - raise SourceLine(obj, "writable", UnsupportedRequirement).makeError("InitialWorkDir feature 'writable: true' not supported") + if obj.get("writable") and self.work_api != "containers": + raise SourceLine(obj, "writable", UnsupportedRequirement).makeError("InitialWorkDir feature 'writable: true' not supported with --api=jobs") if obj.get("class") == "DockerRequirement": if obj.get("dockerOutputDirectory"): - # TODO: can be supported by containers API, but not jobs API. - raise SourceLine(obj, "dockerOutputDirectory", UnsupportedRequirement).makeError( - "Option 'dockerOutputDirectory' of DockerRequirement not supported.") + if self.work_api != "containers": + raise SourceLine(obj, "dockerOutputDirectory", UnsupportedRequirement).makeError( + "Option 'dockerOutputDirectory' of DockerRequirement not supported with --api=jobs.") + if not obj.get("dockerOutputDirectory").startswith('/'): + raise SourceLine(obj, "dockerOutputDirectory", validate.ValidationException).makeError( + "Option 'dockerOutputDirectory' must be an absolute path.") for v in obj.itervalues(): self.check_features(v) elif isinstance(obj, list): @@ -281,7 +285,7 @@ class ArvCwlRunner(object): def rewrite(fileobj): fileobj["location"] = generatemapper.mapper(fileobj["location"]).target - for k in ("basename", "listing", "contents", "nameext", "nameroot", "dirname"): + for k in ("listing", "contents", "nameext", "nameroot", "dirname"): if k in fileobj: del fileobj[k] @@ -350,7 +354,7 @@ class ArvCwlRunner(object): make_fs_access = kwargs.get("make_fs_access") or partial(CollectionFsAccess, collection_cache=self.collection_cache) self.fs_access = make_fs_access(kwargs["basedir"]) - + self.secret_store = kwargs.get("secret_store") self.trash_intermediate = kwargs["trash_intermediate"] if self.trash_intermediate and self.work_api != "containers": @@ -367,8 +371,7 @@ class ArvCwlRunner(object): # Upload direct dependencies of workflow steps, get back mapping of files to keep references. # Also uploads docker images. - override_tools = {} - upload_workflow_deps(self, tool, override_tools) + merged_map = upload_workflow_deps(self, tool) # Reload tool object which may have been updated by # upload_workflow_deps @@ -376,8 +379,7 @@ class ArvCwlRunner(object): makeTool=self.arv_make_tool, loader=tool.doc_loader, avsc_names=tool.doc_schema, - metadata=tool.metadata, - override_tools=override_tools) + metadata=tool.metadata) # Upload local file references in the job order. job_order = upload_job_order(self, "%s input" % kwargs["name"], @@ -391,7 +393,8 @@ class ArvCwlRunner(object): kwargs.get("enable_reuse"), uuid=existing_uuid, submit_runner_ram=kwargs.get("submit_runner_ram"), - name=kwargs["name"]) + name=kwargs["name"], + merged_map=merged_map) tmpl.save() # cwltool.main will write our return value to stdout. return (tmpl.uuid, "success") @@ -400,10 +403,12 @@ class ArvCwlRunner(object): self.project_uuid, uuid=existing_uuid, submit_runner_ram=kwargs.get("submit_runner_ram"), - name=kwargs["name"]), + name=kwargs["name"], + merged_map=merged_map), "success") self.ignore_docker_for_reuse = kwargs.get("ignore_docker_for_reuse") + self.eval_timeout = kwargs.get("eval_timeout") kwargs["make_fs_access"] = make_fs_access kwargs["enable_reuse"] = kwargs.get("enable_reuse") @@ -412,15 +417,22 @@ class ArvCwlRunner(object): kwargs["compute_checksum"] = kwargs.get("compute_checksum") if self.work_api == "containers": + if self.ignore_docker_for_reuse: + raise Exception("--ignore-docker-for-reuse not supported with containers API.") kwargs["outdir"] = "/var/spool/cwl" kwargs["docker_outdir"] = "/var/spool/cwl" kwargs["tmpdir"] = "/tmp" kwargs["docker_tmpdir"] = "/tmp" elif self.work_api == "jobs": + if kwargs["priority"] != DEFAULT_PRIORITY: + raise Exception("--priority not implemented for jobs API.") kwargs["outdir"] = "$(task.outdir)" kwargs["docker_outdir"] = "$(task.outdir)" kwargs["tmpdir"] = "$(task.tmpdir)" + if kwargs["priority"] < 1 or kwargs["priority"] > 1000: + raise Exception("--priority must be in the range 1..1000.") + runnerjob = None if kwargs.get("submit"): # Submit a runner job to run the workflow for us. @@ -438,7 +450,10 @@ class ArvCwlRunner(object): name=kwargs.get("name"), on_error=kwargs.get("on_error"), submit_runner_image=kwargs.get("submit_runner_image"), - intermediate_output_ttl=kwargs.get("intermediate_output_ttl")) + intermediate_output_ttl=kwargs.get("intermediate_output_ttl"), + merged_map=merged_map, + priority=kwargs.get("priority"), + secret_store=kwargs.get("secret_store")) elif self.work_api == "jobs": runnerjob = RunnerJob(self, tool, job_order, kwargs.get("enable_reuse"), self.output_name, @@ -446,7 +461,8 @@ class ArvCwlRunner(object): submit_runner_ram=kwargs.get("submit_runner_ram"), name=kwargs.get("name"), on_error=kwargs.get("on_error"), - submit_runner_image=kwargs.get("submit_runner_image")) + submit_runner_image=kwargs.get("submit_runner_image"), + merged_map=merged_map) elif "cwl_runner_job" not in kwargs and self.work_api == "jobs": # Create pipeline for local run self.pipeline = self.api.pipeline_instances().create( @@ -554,7 +570,7 @@ def versionstring(): arvpkg = pkg_resources.require("arvados-python-client") cwlpkg = pkg_resources.require("cwltool") - return "%s %s %s, %s %s, %s %s" % (sys.argv[0], __version__, arvcwlpkg[0].version, + return "%s %s, %s %s, %s %s" % (sys.argv[0], arvcwlpkg[0].version, "arvados-python-client", arvpkg[0].version, "cwltool", cwlpkg[0].version) @@ -575,7 +591,7 @@ def arg_parser(): # type: () -> argparse.ArgumentParser exgroup = parser.add_mutually_exclusive_group() exgroup.add_argument("--print-dot", action="store_true", help="Print workflow visualization in graphviz format and exit") - exgroup.add_argument("--version", action="store_true", help="Print version and exit") + exgroup.add_argument("--version", action="version", help="Print version and exit", version=versionstring()) exgroup.add_argument("--validate", action="store_true", help="Validate CWL document only.") exgroup = parser.add_mutually_exclusive_group() @@ -657,6 +673,10 @@ def arg_parser(): # type: () -> argparse.ArgumentParser help="If N > 0, intermediate output collections will be trashed N seconds after creation. Default is 0 (don't trash).", default=0) + parser.add_argument("--priority", type=int, + help="Workflow priority (range 1..1000, higher has precedence over lower, containers api only)", + default=DEFAULT_PRIORITY) + exgroup = parser.add_mutually_exclusive_group() exgroup.add_argument("--trash-intermediate", action="store_true", default=False, dest="trash_intermediate", @@ -665,14 +685,14 @@ def arg_parser(): # type: () -> argparse.ArgumentParser default=False, dest="trash_intermediate", help="Do not trash intermediate outputs (default).") - parser.add_argument("workflow", type=str, nargs="?", default=None, help="The workflow to execute") + parser.add_argument("workflow", type=str, default=None, help="The workflow to execute") parser.add_argument("job_order", nargs=argparse.REMAINDER, help="The input object to the workflow.") return parser def add_arv_hints(): - cwltool.draft2tool.ACCEPTLIST_EN_RELAXED_RE = re.compile(r".*") - cwltool.draft2tool.ACCEPTLIST_RE = cwltool.draft2tool.ACCEPTLIST_EN_RELAXED_RE + cwltool.command_line_tool.ACCEPTLIST_EN_RELAXED_RE = re.compile(r".*") + cwltool.command_line_tool.ACCEPTLIST_RE = cwltool.command_line_tool.ACCEPTLIST_EN_RELAXED_RE res = pkg_resources.resource_stream(__name__, 'arv-cwl-schema.yml') use_custom_schema("v1.0", "http://arvados.org/cwl", res.read()) res.close() @@ -693,10 +713,6 @@ def main(args, stdout, stderr, api_client=None, keep_client=None): job_order_object = None arvargs = parser.parse_args(args) - if arvargs.version: - print versionstring() - return - if arvargs.update_workflow: if arvargs.update_workflow.find('-7fd4e-') == 5: want_api = 'containers'