X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/2cf1740bfe7e5790752654ce1b7e0220bc194004..d825b0330a1b51d8ccbb25e7dc7d9aac26e781e0:/sdk/cwl/arvados_cwl/arvworkflow.py diff --git a/sdk/cwl/arvados_cwl/arvworkflow.py b/sdk/cwl/arvados_cwl/arvworkflow.py index b26bee8247..02c9c7a97a 100644 --- a/sdk/cwl/arvados_cwl/arvworkflow.py +++ b/sdk/cwl/arvados_cwl/arvworkflow.py @@ -2,27 +2,33 @@ # # SPDX-License-Identifier: Apache-2.0 +from past.builtins import basestring +from future.utils import viewitems + import os import json import copy import logging from schema_salad.sourceline import SourceLine, cmap +import schema_salad.ref_resolver from cwltool.pack import pack -from cwltool.load_tool import fetch_document +from cwltool.load_tool import fetch_document, resolve_and_validate_document from cwltool.process import shortname from cwltool.workflow import Workflow, WorkflowException, WorkflowStep -from cwltool.pathmapper import adjustFileObjs, adjustDirObjs, visit_class -from cwltool.builder import Builder +from cwltool.utils import adjustFileObjs, adjustDirObjs, visit_class, normalizeFilesDirs from cwltool.context import LoadingContext import ruamel.yaml as yaml from .runner import (upload_dependencies, packed_workflow, upload_workflow_collection, - trim_anonymous_location, remove_redundant_fields, discover_secondary_files) + trim_anonymous_location, remove_redundant_fields, discover_secondary_files, + make_builder, arvados_jobs_image) from .pathmapper import ArvPathMapper, trim_listing -from .arvtool import ArvadosCommandTool, check_cluster_target +from .arvtool import ArvadosCommandTool, set_cluster_target +from ._version import __version__ + from .perf import Perf logger = logging.getLogger('arvados.cwl-runner') @@ -31,10 +37,13 @@ metrics = logging.getLogger('arvados.cwl-runner.metrics') max_res_pars = ("coresMin", "coresMax", "ramMin", "ramMax", "tmpdirMin", "tmpdirMax") sum_res_pars = ("outdirMin", "outdirMax") -def upload_workflow(arvRunner, tool, job_order, project_uuid, uuid=None, - submit_runner_ram=0, name=None, merged_map=None): +def upload_workflow(arvRunner, tool, job_order, project_uuid, + runtimeContext, uuid=None, + submit_runner_ram=0, name=None, merged_map=None, + submit_runner_image=None, + git_info=None): - packed = packed_workflow(arvRunner, tool, merged_map) + packed = packed_workflow(arvRunner, tool, merged_map, runtimeContext, git_info) adjustDirObjs(job_order, trim_listing) adjustFileObjs(job_order, trim_anonymous_location) @@ -50,20 +59,30 @@ def upload_workflow(arvRunner, tool, job_order, project_uuid, uuid=None, name = tool.tool.get("label", os.path.basename(tool.tool["id"])) upload_dependencies(arvRunner, name, tool.doc_loader, - packed, tool.tool["id"], False) + packed, tool.tool["id"], False, + runtimeContext) + + wf_runner_resources = None + + hints = main.get("hints", []) + found = False + for h in hints: + if h["class"] == "http://arvados.org/cwl#WorkflowRunnerResources": + wf_runner_resources = h + found = True + break + if not found: + wf_runner_resources = {"class": "http://arvados.org/cwl#WorkflowRunnerResources"} + hints.append(wf_runner_resources) + + wf_runner_resources["acrContainerImage"] = arvados_jobs_image(arvRunner, + submit_runner_image or "arvados/jobs:"+__version__, + runtimeContext) if submit_runner_ram: - hints = main.get("hints", []) - found = False - for h in hints: - if h["class"] == "http://arvados.org/cwl#WorkflowRunnerResources": - h["ramMin"] = submit_runner_ram - found = True - break - if not found: - hints.append({"class": "http://arvados.org/cwl#WorkflowRunnerResources", - "ramMin": submit_runner_ram}) - main["hints"] = hints + wf_runner_resources["ramMin"] = submit_runner_ram + + main["hints"] = hints body = { "workflow": { @@ -128,15 +147,25 @@ class ArvadosWorkflowStep(WorkflowStep): **argv ): # type: (...) -> None - super(ArvadosWorkflowStep, self).__init__(toolpath_object, pos, loadingContext, *argc, **argv) - self.tool["class"] = "WorkflowStep" + if arvrunner.fast_submit: + self.tool = toolpath_object + self.tool["inputs"] = [] + self.tool["outputs"] = [] + else: + super(ArvadosWorkflowStep, self).__init__(toolpath_object, pos, loadingContext, *argc, **argv) + self.tool["class"] = "WorkflowStep" self.arvrunner = arvrunner def job(self, joborder, output_callback, runtimeContext): - builder = self._init_job({shortname(k): v for k,v in joborder.items()}, runtimeContext) - check_cluster_target(self, builder, runtimeContext) + runtimeContext = runtimeContext.copy() + runtimeContext.toplevel = True # Preserve behavior for #13365 + + builder = make_builder({shortname(k): v for k,v in viewitems(joborder)}, self.hints, self.requirements, + runtimeContext, self.metadata) + runtimeContext = set_cluster_target(self.tool, self.arvrunner, builder, runtimeContext) return super(ArvadosWorkflowStep, self).job(joborder, output_callback, runtimeContext) + class ArvadosWorkflow(Workflow): """Wrap cwltool Workflow to override selected methods.""" @@ -148,11 +177,12 @@ class ArvadosWorkflow(Workflow): self.wf_reffiles = [] self.loadingContext = loadingContext super(ArvadosWorkflow, self).__init__(toolpath_object, loadingContext) + self.cluster_target_req, _ = self.get_requirement("http://arvados.org/cwl#ClusterTarget") def job(self, joborder, output_callback, runtimeContext): - builder = self._init_job(joborder, runtimeContext) - check_cluster_target(self, builder, runtimeContext) + builder = make_builder(joborder, self.hints, self.requirements, runtimeContext, self.metadata) + runtimeContext = set_cluster_target(self.tool, self.arvrunner, builder, runtimeContext) req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer") if not req: @@ -163,25 +193,31 @@ class ArvadosWorkflow(Workflow): with SourceLine(self.tool, None, WorkflowException, logger.isEnabledFor(logging.DEBUG)): if "id" not in self.tool: raise WorkflowException("%s object must have 'id'" % (self.tool["class"])) - document_loader, workflowobj, uri = (self.doc_loader, self.doc_loader.fetch(self.tool["id"]), self.tool["id"]) - discover_secondary_files(self.tool["inputs"], joborder) + discover_secondary_files(self.arvrunner.fs_access, builder, + self.tool["inputs"], joborder) + normalizeFilesDirs(joborder) with Perf(metrics, "subworkflow upload_deps"): upload_dependencies(self.arvrunner, os.path.basename(joborder.get("id", "#")), - document_loader, + self.doc_loader, joborder, joborder.get("id", "#"), - False) + False, + runtimeContext) if self.wf_pdh is None: - workflowobj["requirements"] = dedup_reqs(self.requirements) - workflowobj["hints"] = dedup_reqs(self.hints) + packed = pack(self.loadingContext, self.tool["id"], loader=self.doc_loader) - packed = pack(document_loader, workflowobj, uri, self.metadata) + for p in packed["$graph"]: + if p["id"] == "#main": + p["requirements"] = dedup_reqs(self.requirements) + p["hints"] = dedup_reqs(self.hints) def visit(item): + if "requirements" in item: + item["requirements"] = [i for i in item["requirements"] if i["class"] != "DockerRequirement"] for t in ("hints", "requirements"): if t not in item: continue @@ -209,10 +245,11 @@ class ArvadosWorkflow(Workflow): upload_dependencies(self.arvrunner, runtimeContext.name, - document_loader, + self.doc_loader, packed, - uri, - False) + self.tool["id"], + False, + runtimeContext) # Discover files/directories referenced by the # workflow (mainly "default" values) @@ -276,7 +313,21 @@ class ArvadosWorkflow(Workflow): if self.wf_pdh is None: adjustFileObjs(packed, keepmount) adjustDirObjs(packed, keepmount) - self.wf_pdh = upload_workflow_collection(self.arvrunner, shortname(self.tool["id"]), packed) + self.wf_pdh = upload_workflow_collection(self.arvrunner, shortname(self.tool["id"]), packed, runtimeContext) + + self.loadingContext = self.loadingContext.copy() + self.loadingContext.metadata = self.loadingContext.metadata.copy() + self.loadingContext.metadata["http://commonwl.org/cwltool#original_cwlVersion"] = "v1.0" + + if len(job_res_reqs) == 1: + # RAM request needs to be at least 128 MiB or the workflow + # runner itself won't run reliably. + if job_res_reqs[0].get("ramMin", 1024) < 128: + job_res_reqs[0]["ramMin"] = 128 + + arguments = ["--no-container", "--move-outputs", "--preserve-entire-environment", "workflow.cwl", "cwl.input.yml"] + if runtimeContext.debug: + arguments.insert(0, '--debug') wf_runner = cmap({ "class": "CommandLineTool", @@ -297,7 +348,7 @@ class ArvadosWorkflow(Workflow): }] }], "hints": self.hints, - "arguments": ["--no-container", "--move-outputs", "--preserve-entire-environment", "workflow.cwl#main", "cwl.input.yml"], + "arguments": arguments, "id": "#" }) return ArvadosCommandTool(self.arvrunner, wf_runner, self.loadingContext).job(joborder_resolved, output_callback, runtimeContext)