X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/eae48c31bb338689ec67fbc6a14a2e0b1fb5e3b6..495a7df016f63ae348000b1c046f8e1c5f1c5903:/sdk/cwl/arvados_cwl/runner.py diff --git a/sdk/cwl/arvados_cwl/runner.py b/sdk/cwl/arvados_cwl/runner.py index a1142544f5..1c3625e26b 100644 --- a/sdk/cwl/arvados_cwl/runner.py +++ b/sdk/cwl/arvados_cwl/runner.py @@ -6,6 +6,8 @@ import json import re from cStringIO import StringIO +from schema_salad.sourceline import SourceLine + import cwltool.draft2tool from cwltool.draft2tool import CommandLineTool import cwltool.workflow @@ -21,6 +23,7 @@ import ruamel.yaml as yaml from .arvdocker import arv_docker_get_image from .pathmapper import ArvPathMapper from ._version import __version__ +from . import done logger = logging.getLogger('arvados.cwl-runner') @@ -57,7 +60,7 @@ def upload_dependencies(arvrunner, name, document_loader, loaded = set() def loadref(b, u): - joined = urlparse.urljoin(b, u) + joined = document_loader.fetcher.urljoin(b, u) defrg, _ = urlparse.urldefrag(joined) if defrg not in loaded: loaded.add(defrg) @@ -85,7 +88,7 @@ def upload_dependencies(arvrunner, name, document_loader, sc = scandeps(uri, scanobj, loadref_fields, set(("$include", "$schemas", "location")), - loadref) + loadref, urljoin=document_loader.fetcher.urljoin) normalizeFilesDirs(sc) @@ -112,7 +115,8 @@ def upload_docker(arvrunner, tool): if docker_req: if docker_req.get("dockerOutputDirectory"): # TODO: can be supported by containers API, but not jobs API. - raise UnsupportedRequirement("Option 'dockerOutputDirectory' of DockerRequirement not supported.") + raise SourceLine(docker_req, "dockerOutputDirectory", UnsupportedRequirement).makeError( + "Option 'dockerOutputDirectory' of DockerRequirement not supported.") arv_docker_get_image(arvrunner.api, docker_req, True, arvrunner.project_uuid) elif isinstance(tool, cwltool.workflow.Workflow): for s in tool.steps: @@ -161,7 +165,9 @@ def arvados_jobs_image(arvrunner): return img class Runner(object): - def __init__(self, runner, tool, job_order, enable_reuse, output_name): + def __init__(self, runner, tool, job_order, enable_reuse, + output_name, output_tags, submit_runner_ram=0, + name=None, on_error=None): self.arvrunner = runner self.tool = tool self.job_order = job_order @@ -170,48 +176,78 @@ class Runner(object): self.uuid = None self.final_output = None self.output_name = output_name + self.output_tags = output_tags + self.name = name + self.on_error = on_error + + if submit_runner_ram: + self.submit_runner_ram = submit_runner_ram + else: + self.submit_runner_ram = 1024 + + if self.submit_runner_ram <= 0: + raise Exception("Value of --submit-runner-ram must be greater than zero") def update_pipeline_component(self, record): pass def arvados_job_spec(self, *args, **kwargs): - self.name = os.path.basename(self.tool.tool["id"]) + if self.name is None: + self.name = self.tool.tool.get("label") or os.path.basename(self.tool.tool["id"]) + + # Need to filter this out, gets added by cwltool when providing + # parameters on the command line. + if "job_order" in self.job_order: + del self.job_order["job_order"] + workflowmapper = upload_instance(self.arvrunner, self.name, self.tool, self.job_order) adjustDirObjs(self.job_order, trim_listing) return workflowmapper def done(self, record): - if record["state"] == "Complete": - if record.get("exit_code") is not None: - if record["exit_code"] == 33: - processStatus = "UnsupportedRequirement" - elif record["exit_code"] == 0: - processStatus = "success" + try: + if record["state"] == "Complete": + if record.get("exit_code") is not None: + if record["exit_code"] == 33: + processStatus = "UnsupportedRequirement" + elif record["exit_code"] == 0: + processStatus = "success" + else: + processStatus = "permanentFail" else: - processStatus = "permanentFail" + processStatus = "success" else: - processStatus = "success" - else: - processStatus = "permanentFail" + processStatus = "permanentFail" - outputs = None - try: - try: - self.final_output = record["output"] - outc = arvados.collection.CollectionReader(self.final_output, + outputs = {} + + if processStatus == "permanentFail": + logc = arvados.collection.CollectionReader(record["log"], api_client=self.arvrunner.api, keep_client=self.arvrunner.keep_client, num_retries=self.arvrunner.num_retries) + done.logtail(logc, logger, "%s error log:" % self.arvrunner.label(self), maxlen=40) + + self.final_output = record["output"] + outc = arvados.collection.CollectionReader(self.final_output, + api_client=self.arvrunner.api, + keep_client=self.arvrunner.keep_client, + num_retries=self.arvrunner.num_retries) + if "cwl.output.json" in outc: with outc.open("cwl.output.json") as f: - outputs = json.load(f) - def keepify(fileobj): - path = fileobj["location"] - if not path.startswith("keep:"): - fileobj["location"] = "keep:%s/%s" % (record["output"], path) - adjustFileObjs(outputs, keepify) - adjustDirObjs(outputs, keepify) - except Exception as e: - logger.error("While getting final output object: %s", e) + if f.size() > 0: + outputs = json.load(f) + def keepify(fileobj): + path = fileobj["location"] + if not path.startswith("keep:"): + fileobj["location"] = "keep:%s/%s" % (record["output"], path) + adjustFileObjs(outputs, keepify) + adjustDirObjs(outputs, keepify) + except Exception as e: + logger.exception("[%s] While getting final output object: %s", self.name, e) + self.arvrunner.output_callback({}, "permanentFail") + else: self.arvrunner.output_callback(outputs, processStatus) finally: - del self.arvrunner.processes[record["uuid"]] + if record["uuid"] in self.arvrunner.processes: + del self.arvrunner.processes[record["uuid"]]