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