X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/b6e15a581be2e5b2387fe18ecb01714fbc21d3aa..060d38d627bd1e51dd2b3c6e7de9af6aa7d7b6f3:/sdk/cwl/arvados_cwl/arvjob.py diff --git a/sdk/cwl/arvados_cwl/arvjob.py b/sdk/cwl/arvados_cwl/arvjob.py index 4bbbda26fb..64cd2aa04e 100644 --- a/sdk/cwl/arvados_cwl/arvjob.py +++ b/sdk/cwl/arvados_cwl/arvjob.py @@ -1,20 +1,30 @@ +# Copyright (C) The Arvados Authors. All rights reserved. +# +# SPDX-License-Identifier: Apache-2.0 + import logging import re import copy import json import time -from cwltool.process import get_feature, shortname +from cwltool.process import get_feature, shortname, UnsupportedRequirement from cwltool.errors import WorkflowException from cwltool.draft2tool import revmap_file, CommandLineTool from cwltool.load_tool import fetch_document from cwltool.builder import Builder +from cwltool.pathmapper import adjustFileObjs, adjustDirObjs + +from schema_salad.sourceline import SourceLine + +import ruamel.yaml as yaml import arvados.collection +from arvados.errors import ApiError from .arvdocker import arv_docker_get_image -from .runner import Runner, arvados_jobs_image -from .pathmapper import InitialWorkDirPathMapper +from .runner import Runner, arvados_jobs_image, packed_workflow, upload_workflow_collection, trim_anonymous_location +from .pathmapper import VwdPathMapper, trim_listing from .perf import Perf from . import done from ._version import __version__ @@ -46,8 +56,8 @@ class ArvadosJob(object): keep_client=self.arvrunner.keep_client, num_retries=self.arvrunner.num_retries) script_parameters["task.vwd"] = {} - generatemapper = InitialWorkDirPathMapper([self.generatefiles], "", "", - separateDirs=False) + generatemapper = VwdPathMapper([self.generatefiles], "", "", + separateDirs=False) with Perf(metrics, "createfiles %s" % self.name): for f, p in generatemapper.items(): @@ -55,8 +65,9 @@ class ArvadosJob(object): with vwd.open(p.target, "w") as n: n.write(p.resolved.encode("utf-8")) - with Perf(metrics, "generatefiles.save_new %s" % self.name): - vwd.save_new() + if vwd: + with Perf(metrics, "generatefiles.save_new %s" % self.name): + vwd.save_new() for f, p in generatemapper.items(): if p.type == "File": @@ -104,6 +115,7 @@ class ArvadosJob(object): if runtime_req: if "keep_cache" in runtime_req: runtime_constraints["keep_cache_mb_per_task"] = runtime_req["keep_cache"] + runtime_constraints["min_ram_mb_per_node"] += runtime_req["keep_cache"] if "outputDirType" in runtime_req: if runtime_req["outputDirType"] == "local_output_dir": script_parameters["task.keepTmpOutput"] = False @@ -116,6 +128,12 @@ class ArvadosJob(object): if not self.arvrunner.ignore_docker_for_reuse: filters.append(["docker_image_locator", "in docker", runtime_constraints["docker_image"]]) + enable_reuse = kwargs.get("enable_reuse", True) + if enable_reuse: + reuse_req, _ = get_feature(self, "http://arvados.org/cwl#ReuseRequirement") + if reuse_req: + enable_reuse = reuse_req["enableReuse"] + try: with Perf(metrics, "create %s" % self.name): response = self.arvrunner.api.jobs().create( @@ -129,18 +147,35 @@ class ArvadosJob(object): "runtime_constraints": runtime_constraints }, filters=filters, - find_or_create=kwargs.get("enable_reuse", True) + find_or_create=enable_reuse ).execute(num_retries=self.arvrunner.num_retries) self.arvrunner.processes[response["uuid"]] = self self.update_pipeline_component(response) - logger.info("%s %s is %s", self.arvrunner.label(self), response["uuid"], response["state"]) + if response["state"] == "Complete": + logger.info("%s reused job %s", self.arvrunner.label(self), response["uuid"]) + # Give read permission to the desired project on reused jobs + if response["owner_uuid"] != self.arvrunner.project_uuid: + try: + self.arvrunner.api.links().create(body={ + 'link_class': 'permission', + 'name': 'can_read', + 'tail_uuid': self.arvrunner.project_uuid, + 'head_uuid': response["uuid"], + }).execute(num_retries=self.arvrunner.num_retries) + except ApiError as e: + # The user might not have "manage" access on the job: log + # a message and continue. + logger.info("Creating read permission on job %s: %s", + response["uuid"], + e) - if response["state"] in ("Complete", "Failed", "Cancelled"): with Perf(metrics, "done %s" % self.name): self.done(response) + else: + logger.info("%s %s is %s", self.arvrunner.label(self), response["uuid"], response["state"]) except Exception as e: logger.exception("%s error" % (self.arvrunner.label(self))) self.output_callback({}, "permanentFail") @@ -239,9 +274,16 @@ class RunnerJob(Runner): a pipeline template or pipeline instance. """ - workflowmapper = super(RunnerJob, self).arvados_job_spec(dry_run=dry_run, pull_image=pull_image, **kwargs) + if self.tool.tool["id"].startswith("keep:"): + self.job_order["cwl:tool"] = self.tool.tool["id"][5:] + else: + packed = packed_workflow(self.arvrunner, self.tool) + wf_pdh = upload_workflow_collection(self.arvrunner, self.name, packed) + self.job_order["cwl:tool"] = "%s/workflow.cwl#main" % wf_pdh - self.job_order["cwl:tool"] = workflowmapper.mapper(self.tool.tool["id"]).target[5:] + adjustDirObjs(self.job_order, trim_listing) + adjustFileObjs(self.job_order, trim_anonymous_location) + adjustDirObjs(self.job_order, trim_anonymous_location) if self.output_name: self.job_order["arv:output_name"] = self.output_name @@ -355,10 +397,12 @@ class RunnerTemplate(object): if not isinstance(types, list): types = [types] param['required'] = 'null' not in types - non_null_types = set(types) - set(['null']) + non_null_types = [t for t in types if t != "null"] if len(non_null_types) == 1: the_type = [c for c in non_null_types][0] - dataclass = self.type_to_dataclass.get(the_type) + dataclass = None + if isinstance(the_type, basestring): + dataclass = self.type_to_dataclass.get(the_type) if dataclass: param['dataclass'] = dataclass # Note: If we didn't figure out a single appropriate