sdk/cwl/arvados_cwl/__init__.py

   1 #!/usr/bin/env python3
   2 # Copyright (C) The Arvados Authors. All rights reserved.
   3 #
   4 # SPDX-License-Identifier: Apache-2.0
   5
   6 # Implement cwl-runner interface for submitting and running work on Arvados, using
   7 # the Crunch containers API.
   8
   9 from future.utils import viewitems
  10 from builtins import str
  11
  12 import argparse
  13 import logging
  14 import os
  15 import sys
  16 import re
  17 import pkg_resources  # part of setuptools
  18
  19 from schema_salad.sourceline import SourceLine
  20 import schema_salad.validate as validate
  21 import cwltool.main
  22 import cwltool.workflow
  23 import cwltool.process
  24 import cwltool.argparser
  25 from cwltool.errors import WorkflowException
  26 from cwltool.process import shortname, UnsupportedRequirement, use_custom_schema
  27 from cwltool.utils import adjustFileObjs, adjustDirObjs, get_listing
  28
  29 import arvados
  30 import arvados.config
  31 from arvados.keep import KeepClient
  32 from arvados.errors import ApiError
  33 import arvados.commands._util as arv_cmd
  34 from arvados.api import OrderedJsonModel
  35
  36 from .perf import Perf
  37 from ._version import __version__
  38 from .executor import ArvCwlExecutor
  39 from .fsaccess import workflow_uuid_pattern
  40
  41 # These aren't used directly in this file but
  42 # other code expects to import them from here
  43 from .arvcontainer import ArvadosContainer
  44 from .arvtool import ArvadosCommandTool
  45 from .fsaccess import CollectionFsAccess, CollectionCache, CollectionFetcher
  46 from .util import get_current_container
  47 from .executor import RuntimeStatusLoggingHandler, DEFAULT_PRIORITY
  48 from .arvworkflow import ArvadosWorkflow
  49
  50 logger = logging.getLogger('arvados.cwl-runner')
  51 metrics = logging.getLogger('arvados.cwl-runner.metrics')
  52 logger.setLevel(logging.INFO)
  53
  54 arvados.log_handler.setFormatter(logging.Formatter(
  55         '%(asctime)s %(name)s %(levelname)s: %(message)s',
  56         '%Y-%m-%d %H:%M:%S'))
  57
  58 def versionstring():
  59     """Print version string of key packages for provenance and debugging."""
  60
  61     arvcwlpkg = pkg_resources.require("arvados-cwl-runner")
  62     arvpkg = pkg_resources.require("arvados-python-client")
  63     cwlpkg = pkg_resources.require("cwltool")
  64
  65     return "%s %s, %s %s, %s %s" % (sys.argv[0], arvcwlpkg[0].version,
  66                                     "arvados-python-client", arvpkg[0].version,
  67                                     "cwltool", cwlpkg[0].version)
  68
  69
  70 def arg_parser():  # type: () -> argparse.ArgumentParser
  71     parser = argparse.ArgumentParser(description='Arvados executor for Common Workflow Language')
  72
  73     parser.add_argument("--basedir",
  74                         help="Base directory used to resolve relative references in the input, default to directory of input object file or current directory (if inputs piped/provided on command line).")
  75     parser.add_argument("--outdir", default=os.path.abspath('.'),
  76                         help="Output directory, default current directory")
  77
  78     parser.add_argument("--eval-timeout",
  79                         help="Time to wait for a Javascript expression to evaluate before giving an error, default 20s.",
  80                         type=float,
  81                         default=20)
  82
  83     exgroup = parser.add_mutually_exclusive_group()
  84     exgroup.add_argument("--print-dot", action="store_true",
  85                          help="Print workflow visualization in graphviz format and exit")
  86     exgroup.add_argument("--version", action="version", help="Print version and exit", version=versionstring())
  87     exgroup.add_argument("--validate", action="store_true", help="Validate CWL document only.")
  88
  89     exgroup = parser.add_mutually_exclusive_group()
  90     exgroup.add_argument("--verbose", action="store_true", help="Default logging")
  91     exgroup.add_argument("--quiet", action="store_true", help="Only print warnings and errors.")
  92     exgroup.add_argument("--debug", action="store_true", help="Print even more logging")
  93
  94     parser.add_argument("--metrics", action="store_true", help="Print timing metrics")
  95
  96     parser.add_argument("--tool-help", action="store_true", help="Print command line help for tool")
  97
  98     exgroup = parser.add_mutually_exclusive_group()
  99     exgroup.add_argument("--enable-reuse", action="store_true",
 100                         default=True, dest="enable_reuse",
 101                         help="Enable container reuse (default)")
 102     exgroup.add_argument("--disable-reuse", action="store_false",
 103                         default=True, dest="enable_reuse",
 104                         help="Disable container reuse")
 105
 106     parser.add_argument("--project-uuid", metavar="UUID", help="Project that will own the workflow containers, if not provided, will go to home project.")
 107     parser.add_argument("--output-name", help="Name to use for collection that stores the final output.", default=None)
 108     parser.add_argument("--output-tags", help="Tags for the final output collection separated by commas, e.g., '--output-tags tag0,tag1,tag2'.", default=None)
 109     parser.add_argument("--ignore-docker-for-reuse", action="store_true",
 110                         help="Ignore Docker image version when deciding whether to reuse past containers.",
 111                         default=False)
 112
 113     exgroup = parser.add_mutually_exclusive_group()
 114     exgroup.add_argument("--submit", action="store_true", help="Submit workflow to run on Arvados.",
 115                         default=True, dest="submit")
 116     exgroup.add_argument("--local", action="store_false", help="Run workflow on local host (submits containers to Arvados).",
 117                         default=True, dest="submit")
 118     exgroup.add_argument("--create-template", action="store_true", help="(Deprecated) synonym for --create-workflow.",
 119                          dest="create_workflow")
 120     exgroup.add_argument("--create-workflow", action="store_true", help="Register an Arvados workflow that can be run from Workbench")
 121     exgroup.add_argument("--update-workflow", metavar="UUID", help="Update an existing Arvados workflow with the given UUID.")
 122
 123     exgroup = parser.add_mutually_exclusive_group()
 124     exgroup.add_argument("--wait", action="store_true", help="After submitting workflow runner, wait for completion.",
 125                         default=True, dest="wait")
 126     exgroup.add_argument("--no-wait", action="store_false", help="Submit workflow runner and exit.",
 127                         default=True, dest="wait")
 128
 129     exgroup = parser.add_mutually_exclusive_group()
 130     exgroup.add_argument("--log-timestamps", action="store_true", help="Prefix logging lines with timestamp",
 131                         default=True, dest="log_timestamps")
 132     exgroup.add_argument("--no-log-timestamps", action="store_false", help="No timestamp on logging lines",
 133                         default=True, dest="log_timestamps")
 134
 135     parser.add_argument("--api",
 136                         default=None, dest="work_api",
 137                         choices=("containers",),
 138                         help="Select work submission API.  Only supports 'containers'")
 139
 140     parser.add_argument("--compute-checksum", action="store_true", default=False,
 141                         help="Compute checksum of contents while collecting outputs",
 142                         dest="compute_checksum")
 143
 144     parser.add_argument("--submit-runner-ram", type=int,
 145                         help="RAM (in MiB) required for the workflow runner job (default 1024)",
 146                         default=None)
 147
 148     parser.add_argument("--submit-runner-image",
 149                         help="Docker image for workflow runner job, default arvados/jobs:%s" % __version__,
 150                         default=None)
 151
 152     parser.add_argument("--always-submit-runner", action="store_true",
 153                         help="When invoked with --submit --wait, always submit a runner to manage the workflow, even when only running a single CommandLineTool",
 154                         default=False)
 155
 156     parser.add_argument("--match-submitter-images", action="store_true",
 157                         default=False, dest="match_local_docker",
 158                         help="Where Arvados has more than one Docker image of the same name, use image from the Docker instance on the submitting node.")
 159
 160     exgroup = parser.add_mutually_exclusive_group()
 161     exgroup.add_argument("--submit-request-uuid",
 162                          default=None,
 163                          help="Update and commit to supplied container request instead of creating a new one.",
 164                          metavar="UUID")
 165     exgroup.add_argument("--submit-runner-cluster",
 166                          help="Submit workflow runner to a remote cluster",
 167                          default=None,
 168                          metavar="CLUSTER_ID")
 169
 170     parser.add_argument("--collection-cache-size", type=int,
 171                         default=None,
 172                         help="Collection cache size (in MiB, default 256).")
 173
 174     parser.add_argument("--name",
 175                         help="Name to use for workflow execution instance.",
 176                         default=None)
 177
 178     parser.add_argument("--on-error",
 179                         help="Desired workflow behavior when a step fails.  One of 'stop' (do not submit any more steps) or "
 180                         "'continue' (may submit other steps that are not downstream from the error). Default is 'continue'.",
 181                         default="continue", choices=("stop", "continue"))
 182
 183     parser.add_argument("--enable-dev", action="store_true",
 184                         help="Enable loading and running development versions "
 185                              "of the CWL standards.", default=False)
 186     parser.add_argument('--storage-classes', default="default",
 187                         help="Specify comma separated list of storage classes to be used when saving final workflow output to Keep.")
 188     parser.add_argument('--intermediate-storage-classes', default="default",
 189                         help="Specify comma separated list of storage classes to be used when saving intermediate workflow output to Keep.")
 190
 191     parser.add_argument("--intermediate-output-ttl", type=int, metavar="N",
 192                         help="If N > 0, intermediate output collections will be trashed N seconds after creation.  Default is 0 (don't trash).",
 193                         default=0)
 194
 195     parser.add_argument("--priority", type=int,
 196                         help="Workflow priority (range 1..1000, higher has precedence over lower)",
 197                         default=DEFAULT_PRIORITY)
 198
 199     parser.add_argument("--disable-validate", dest="do_validate",
 200                         action="store_false", default=True,
 201                         help=argparse.SUPPRESS)
 202
 203     parser.add_argument("--disable-git", dest="git_info",
 204                         action="store_false", default=True,
 205                         help=argparse.SUPPRESS)
 206
 207     parser.add_argument("--disable-color", dest="enable_color",
 208                         action="store_false", default=True,
 209                         help=argparse.SUPPRESS)
 210
 211     parser.add_argument("--disable-js-validation",
 212                         action="store_true", default=False,
 213                         help=argparse.SUPPRESS)
 214
 215     parser.add_argument("--thread-count", type=int,
 216                         default=0, help="Number of threads to use for job submit and output collection.")
 217
 218     parser.add_argument("--http-timeout", type=int,
 219                         default=5*60, dest="http_timeout", help="API request timeout in seconds. Default is 300 seconds (5 minutes).")
 220
 221     parser.add_argument("--defer-downloads", action="store_true", default=False,
 222                         help="When submitting a workflow, defer downloading HTTP URLs to workflow launch instead of downloading to Keep before submit.")
 223
 224     parser.add_argument("--varying-url-params", type=str, default="",
 225                         help="A comma separated list of URL query parameters that should be ignored when storing HTTP URLs in Keep.")
 226
 227     parser.add_argument("--prefer-cached-downloads", action="store_true", default=False,
 228                         help="If a HTTP URL is found in Keep, skip upstream URL freshness check (will not notice if the upstream has changed, but also not error if upstream is unavailable).")
 229
 230     exgroup = parser.add_mutually_exclusive_group()
 231     exgroup.add_argument("--enable-preemptible", dest="enable_preemptible", default=None, action="store_true", help="Use preemptible instances. Control individual steps with arv:UsePreemptible hint.")
 232     exgroup.add_argument("--disable-preemptible", dest="enable_preemptible", default=None, action="store_false", help="Don't use preemptible instances.")
 233
 234     exgroup = parser.add_mutually_exclusive_group()
 235     exgroup.add_argument("--copy-deps", dest="copy_deps", default=None, action="store_true", help="Copy dependencies into the destination project.")
 236     exgroup.add_argument("--no-copy-deps", dest="copy_deps", default=None, action="store_false", help="Leave dependencies where they are.")
 237
 238     parser.add_argument(
 239         "--skip-schemas",
 240         action="store_true",
 241         help="Skip loading of schemas",
 242         default=False,
 243         dest="skip_schemas",
 244     )
 245
 246     exgroup = parser.add_mutually_exclusive_group()
 247     exgroup.add_argument("--trash-intermediate", action="store_true",
 248                         default=False, dest="trash_intermediate",
 249                          help="Immediately trash intermediate outputs on workflow success.")
 250     exgroup.add_argument("--no-trash-intermediate", action="store_false",
 251                         default=False, dest="trash_intermediate",
 252                         help="Do not trash intermediate outputs (default).")
 253
 254     parser.add_argument("workflow", default=None, help="The workflow to execute")
 255     parser.add_argument("job_order", nargs=argparse.REMAINDER, help="The input object to the workflow.")
 256
 257     return parser
 258
 259 def add_arv_hints():
 260     cwltool.command_line_tool.ACCEPTLIST_EN_RELAXED_RE = re.compile(r".*")
 261     cwltool.command_line_tool.ACCEPTLIST_RE = cwltool.command_line_tool.ACCEPTLIST_EN_RELAXED_RE
 262     supported_versions = ["v1.0", "v1.1", "v1.2"]
 263     for s in supported_versions:
 264         res = pkg_resources.resource_stream(__name__, 'arv-cwl-schema-%s.yml' % s)
 265         customschema = res.read().decode('utf-8')
 266         use_custom_schema(s, "http://arvados.org/cwl", customschema)
 267         res.close()
 268     cwltool.process.supportedProcessRequirements.extend([
 269         "http://arvados.org/cwl#RunInSingleContainer",
 270         "http://arvados.org/cwl#OutputDirType",
 271         "http://arvados.org/cwl#RuntimeConstraints",
 272         "http://arvados.org/cwl#PartitionRequirement",
 273         "http://arvados.org/cwl#APIRequirement",
 274         "http://commonwl.org/cwltool#LoadListingRequirement",
 275         "http://arvados.org/cwl#IntermediateOutput",
 276         "http://arvados.org/cwl#ReuseRequirement",
 277         "http://arvados.org/cwl#ClusterTarget",
 278         "http://arvados.org/cwl#OutputStorageClass",
 279         "http://arvados.org/cwl#ProcessProperties",
 280         "http://commonwl.org/cwltool#CUDARequirement",
 281         "http://arvados.org/cwl#UsePreemptible",
 282         "http://arvados.org/cwl#OutputCollectionProperties",
 283     ])
 284
 285 def exit_signal_handler(sigcode, frame):
 286     logger.error(str(u"Caught signal {}, exiting.").format(sigcode))
 287     sys.exit(-sigcode)
 288
 289 def main(args=sys.argv[1:],
 290          stdout=sys.stdout,
 291          stderr=sys.stderr,
 292          api_client=None,
 293          keep_client=None,
 294          install_sig_handlers=True):
 295     parser = arg_parser()
 296
 297     job_order_object = None
 298     arvargs = parser.parse_args(args)
 299
 300     arvargs.use_container = True
 301     arvargs.relax_path_checks = True
 302     arvargs.print_supported_versions = False
 303
 304     if install_sig_handlers:
 305         arv_cmd.install_signal_handlers()
 306
 307     if arvargs.update_workflow:
 308         if arvargs.update_workflow.find('-7fd4e-') == 5:
 309             want_api = 'containers'
 310         else:
 311             want_api = None
 312         if want_api and arvargs.work_api and want_api != arvargs.work_api:
 313             logger.error(str(u'--update-workflow arg {!r} uses {!r} API, but --api={!r} specified').format(
 314                 arvargs.update_workflow, want_api, arvargs.work_api))
 315             return 1
 316         arvargs.work_api = want_api
 317
 318     if (arvargs.create_workflow or arvargs.update_workflow) and not arvargs.job_order:
 319         job_order_object = ({}, "")
 320
 321     add_arv_hints()
 322
 323     for key, val in viewitems(cwltool.argparser.get_default_args()):
 324         if not hasattr(arvargs, key):
 325             setattr(arvargs, key, val)
 326
 327     try:
 328         if api_client is None:
 329             api_client = arvados.safeapi.ThreadSafeApiCache(
 330                 api_params={"model": OrderedJsonModel(), "timeout": arvargs.http_timeout},
 331                 keep_params={"num_retries": 4})
 332             keep_client = api_client.keep
 333             # Make an API object now so errors are reported early.
 334             api_client.users().current().execute()
 335         if keep_client is None:
 336             keep_client = arvados.keep.KeepClient(api_client=api_client, num_retries=4)
 337         executor = ArvCwlExecutor(api_client, arvargs, keep_client=keep_client, num_retries=4, stdout=stdout)
 338     except WorkflowException as e:
 339         logger.error(e, exc_info=(sys.exc_info()[1] if arvargs.debug else False))
 340         return 1
 341     except Exception:
 342         logger.exception("Error creating the Arvados CWL Executor")
 343         return 1
 344
 345     # Note that unless in debug mode, some stack traces related to user
 346     # workflow errors may be suppressed.
 347     if arvargs.debug:
 348         logger.setLevel(logging.DEBUG)
 349         logging.getLogger('arvados').setLevel(logging.DEBUG)
 350
 351     if arvargs.quiet:
 352         logger.setLevel(logging.WARN)
 353         logging.getLogger('arvados').setLevel(logging.WARN)
 354         logging.getLogger('arvados.arv-run').setLevel(logging.WARN)
 355
 356     if arvargs.metrics:
 357         metrics.setLevel(logging.DEBUG)
 358         logging.getLogger("cwltool.metrics").setLevel(logging.DEBUG)
 359
 360     if arvargs.log_timestamps:
 361         arvados.log_handler.setFormatter(logging.Formatter(
 362             '%(asctime)s %(name)s %(levelname)s: %(message)s',
 363             '%Y-%m-%d %H:%M:%S'))
 364     else:
 365         arvados.log_handler.setFormatter(logging.Formatter('%(name)s %(levelname)s: %(message)s'))
 366
 367     if stdout is sys.stdout:
 368         # cwltool.main has code to work around encoding issues with
 369         # sys.stdout and unix pipes (they default to ASCII encoding,
 370         # we want utf-8), so when stdout is sys.stdout set it to None
 371         # to take advantage of that.  Don't override it for all cases
 372         # since we still want to be able to capture stdout for the
 373         # unit tests.
 374         stdout = None
 375
 376     if arvargs.submit and (arvargs.workflow.startswith("arvwf:") or workflow_uuid_pattern.match(arvargs.workflow)):
 377         executor.loadingContext.do_validate = False
 378         executor.fast_submit = True
 379
 380     return cwltool.main.main(args=arvargs,
 381                              stdout=stdout,
 382                              stderr=stderr,
 383                              executor=executor.arv_executor,
 384                              versionfunc=versionstring,
 385                              job_order_object=job_order_object,
 386                              logger_handler=arvados.log_handler,
 387                              custom_schema_callback=add_arv_hints,
 388                              loadingContext=executor.loadingContext,
 389                              runtimeContext=executor.toplevel_runtimeContext,
 390                              input_required=not (arvargs.create_workflow or arvargs.update_workflow))