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