+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: Apache-2.0
+
import logging
import json
import os
import urllib
+import time
+import datetime
+import ciso8601
import ruamel.yaml as yaml
from cwltool.errors import WorkflowException
from cwltool.process import get_feature, UnsupportedRequirement, shortname
-from cwltool.pathmapper import adjustFiles, adjustDirObjs
+from cwltool.pathmapper import adjustFileObjs, adjustDirObjs, visit_class
from cwltool.utils import aslist
import arvados.collection
from .arvdocker import arv_docker_get_image
from . import done
-from .runner import Runner, arvados_jobs_image, packed_workflow, trim_listing
+from .runner import Runner, arvados_jobs_image, packed_workflow, trim_anonymous_location, remove_redundant_fields
from .fsaccess import CollectionFetcher
-from .pathmapper import NoFollowPathMapper
+from .pathmapper import NoFollowPathMapper, trim_listing
from .perf import Perf
logger = logging.getLogger('arvados.cwl-runner')
"cwd": self.outdir,
"priority": 1,
"state": "Committed",
- "properties": {}
+ "properties": {},
}
runtime_constraints = {}
+
+ resources = self.builder.resources
+ if resources is not None:
+ runtime_constraints["vcpus"] = resources.get("cores", 1)
+ runtime_constraints["ram"] = resources.get("ram") * 2**20
+
mounts = {
self.outdir: {
- "kind": "tmp"
+ "kind": "tmp",
+ "capacity": resources.get("outdirSize", 0) * 2**20
+ },
+ self.tmpdir: {
+ "kind": "tmp",
+ "capacity": resources.get("tmpdirSize", 0) * 2**20
}
}
scheduling_parameters = {}
- dirs = set()
- for f in self.pathmapper.files():
- pdh, p, tp, stg = self.pathmapper.mapper(f)
- if tp == "Directory" and '/' not in pdh:
- mounts[p] = {
- "kind": "collection",
- "portable_data_hash": pdh[5:]
- }
- dirs.add(pdh)
-
- for f in self.pathmapper.files():
- res, p, tp, stg = self.pathmapper.mapper(f)
- if res.startswith("keep:"):
- res = res[5:]
- elif res.startswith("/keep/"):
- res = res[6:]
- else:
+ rf = [self.pathmapper.mapper(f) for f in self.pathmapper.referenced_files]
+ rf.sort(key=lambda k: k.resolved)
+ prevdir = None
+ for resolved, target, tp, stg in rf:
+ if not stg:
continue
- sp = res.split("/", 1)
- pdh = sp[0]
- if pdh not in dirs:
- mounts[p] = {
- "kind": "collection",
- "portable_data_hash": pdh
- }
- if len(sp) == 2:
- mounts[p]["path"] = urllib.unquote(sp[1])
+ if prevdir and target.startswith(prevdir):
+ continue
+ if tp == "Directory":
+ targetdir = target
+ else:
+ targetdir = os.path.dirname(target)
+ sp = resolved.split("/", 1)
+ pdh = sp[0][5:] # remove "keep:"
+ mounts[targetdir] = {
+ "kind": "collection",
+ "portable_data_hash": pdh
+ }
+ if len(sp) == 2:
+ if tp == "Directory":
+ path = sp[1]
+ else:
+ path = os.path.dirname(sp[1])
+ if path and path != "/":
+ mounts[targetdir]["path"] = path
+ prevdir = targetdir + "/"
with Perf(metrics, "generatefiles %s" % self.name):
if self.generatefiles["listing"]:
pull_image,
self.arvrunner.project_uuid)
- resources = self.builder.resources
- if resources is not None:
- runtime_constraints["vcpus"] = resources.get("cores", 1)
- runtime_constraints["ram"] = resources.get("ram") * 2**20
-
api_req, _ = get_feature(self, "http://arvados.org/cwl#APIRequirement")
if api_req:
runtime_constraints["API"] = True
if runtime_req:
if "keep_cache" in runtime_req:
runtime_constraints["keep_cache_ram"] = runtime_req["keep_cache"] * 2**20
+ if "outputDirType" in runtime_req:
+ if runtime_req["outputDirType"] == "local_output_dir":
+ # Currently the default behavior.
+ pass
+ elif runtime_req["outputDirType"] == "keep_output_dir":
+ mounts[self.outdir]= {
+ "kind": "collection",
+ "writable": True
+ }
partition_req, _ = get_feature(self, "http://arvados.org/cwl#PartitionRequirement")
if partition_req:
scheduling_parameters["partitions"] = aslist(partition_req["partition"])
+ intermediate_output_req, _ = get_feature(self, "http://arvados.org/cwl#IntermediateOutput")
+ if intermediate_output_req:
+ self.output_ttl = intermediate_output_req["outputTTL"]
+ else:
+ self.output_ttl = self.arvrunner.intermediate_output_ttl
+
+ if self.output_ttl < 0:
+ raise WorkflowError("Invalid value %d for output_ttl, cannot be less than zero" % container_request["output_ttl"])
+
+ container_request["output_ttl"] = self.output_ttl
container_request["mounts"] = mounts
container_request["runtime_constraints"] = runtime_constraints
- container_request["use_existing"] = kwargs.get("enable_reuse", True)
container_request["scheduling_parameters"] = scheduling_parameters
+ 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"]
+ container_request["use_existing"] = enable_reuse
+
if kwargs.get("runnerjob", "").startswith("arvwf:"):
wfuuid = kwargs["runnerjob"][6:kwargs["runnerjob"].index("#")]
wfrecord = self.arvrunner.api.workflows().get(uuid=wfuuid).execute(num_retries=self.arvrunner.num_retries)
self.output_callback({}, "permanentFail")
def done(self, record):
+ outputs = {}
try:
container = self.arvrunner.api.containers().get(
uuid=record["container_uuid"]
num_retries=self.arvrunner.num_retries)
done.logtail(logc, logger, "%s error log:" % self.arvrunner.label(self))
- outputs = {}
+ if record["output_uuid"]:
+ if self.arvrunner.trash_intermediate or self.arvrunner.intermediate_output_ttl:
+ # Compute the trash time to avoid requesting the collection record.
+ trash_at = ciso8601.parse_datetime_unaware(record["modified_at"]) + datetime.timedelta(0, self.arvrunner.intermediate_output_ttl)
+ aftertime = " at %s" % trash_at.strftime("%Y-%m-%d %H:%M:%S UTC") if self.arvrunner.intermediate_output_ttl else ""
+ orpart = ", or" if self.arvrunner.trash_intermediate and self.arvrunner.intermediate_output_ttl else ""
+ oncomplete = " upon successful completion of the workflow" if self.arvrunner.trash_intermediate else ""
+ logger.info("%s Intermediate output %s (%s) will be trashed%s%s%s." % (
+ self.arvrunner.label(self), record["output_uuid"], container["output"], aftertime, orpart, oncomplete))
+ self.arvrunner.add_intermediate_output(record["output_uuid"])
+
if container["output"]:
outputs = done.done_outputs(self, container, "/tmp", self.outdir, "/keep")
except WorkflowException as e:
"""
adjustDirObjs(self.job_order, trim_listing)
+ visit_class(self.job_order, ("File", "Directory"), trim_anonymous_location)
+ visit_class(self.job_order, ("File", "Directory"), remove_redundant_fields)
container_req = {
"owner_uuid": self.arvrunner.project_uuid,
"ram": 1024*1024 * self.submit_runner_ram,
"API": True
},
+ "use_existing": self.enable_reuse,
"properties": {}
}
if self.on_error:
command.append("--on-error=" + self.on_error)
+ if self.intermediate_output_ttl:
+ command.append("--intermediate-output-ttl=%d" % self.intermediate_output_ttl)
+
+ if self.arvrunner.trash_intermediate:
+ command.append("--trash-intermediate")
+
+ if self.arvrunner.project_uuid:
+ command.append("--project-uuid="+self.arvrunner.project_uuid)
+
command.extend([workflowpath, "/var/lib/cwl/cwl.input.json"])
container_req["command"] = command