sdk/cwl/arvados_cwl/arvdocker.py

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: Apache-2.0
   4
   5 import logging
   6 import sys
   7 import threading
   8 import copy
   9
  10 from schema_salad.sourceline import SourceLine
  11
  12 import cwltool.docker
  13 from cwltool.errors import WorkflowException
  14 import arvados.commands.keepdocker
  15
  16 logger = logging.getLogger('arvados.cwl-runner')
  17
  18 cached_lookups = {}
  19 cached_lookups_lock = threading.Lock()
  20
  21 def arv_docker_get_image(api_client, dockerRequirement, pull_image, project_uuid,
  22                          force_pull, tmp_outdir_prefix):
  23     """Check if a Docker image is available in Keep, if not, upload it using arv-keepdocker."""
  24
  25     if "http://arvados.org/cwl#dockerCollectionPDH" in dockerRequirement:
  26         return dockerRequirement["http://arvados.org/cwl#dockerCollectionPDH"]
  27
  28     if "dockerImageId" not in dockerRequirement and "dockerPull" in dockerRequirement:
  29         dockerRequirement = copy.deepcopy(dockerRequirement)
  30         dockerRequirement["dockerImageId"] = dockerRequirement["dockerPull"]
  31         if hasattr(dockerRequirement, 'lc'):
  32             dockerRequirement.lc.data["dockerImageId"] = dockerRequirement.lc.data["dockerPull"]
  33
  34     global cached_lookups
  35     global cached_lookups_lock
  36     with cached_lookups_lock:
  37         if dockerRequirement["dockerImageId"] in cached_lookups:
  38             return cached_lookups[dockerRequirement["dockerImageId"]]
  39
  40     with SourceLine(dockerRequirement, "dockerImageId", WorkflowException, logger.isEnabledFor(logging.DEBUG)):
  41         sp = dockerRequirement["dockerImageId"].split(":")
  42         image_name = sp[0]
  43         image_tag = sp[1] if len(sp) > 1 else "latest"
  44
  45         images = arvados.commands.keepdocker.list_images_in_arv(api_client, 3,
  46                                                                 image_name=image_name,
  47                                                                 image_tag=image_tag)
  48
  49         if not images:
  50             # Fetch Docker image if necessary.
  51             try:
  52                 cwltool.docker.DockerCommandLineJob.get_image(dockerRequirement, pull_image,
  53                                                               force_pull, tmp_outdir_prefix)
  54             except OSError as e:
  55                 raise WorkflowException("While trying to get Docker image '%s', failed to execute 'docker': %s" % (dockerRequirement["dockerImageId"], e))
  56
  57             # Upload image to Arvados
  58             args = []
  59             if project_uuid:
  60                 args.append("--project-uuid="+project_uuid)
  61             args.append(image_name)
  62             args.append(image_tag)
  63             logger.info("Uploading Docker image %s:%s", image_name, image_tag)
  64             try:
  65                 arvados.commands.put.api_client = api_client
  66                 arvados.commands.keepdocker.main(args, stdout=sys.stderr, install_sig_handlers=False, api=api_client)
  67             except SystemExit as e:
  68                 # If e.code is None or zero, then keepdocker exited normally and we can continue
  69                 if e.code:
  70                     raise WorkflowException("keepdocker exited with code %s" % e.code)
  71
  72             images = arvados.commands.keepdocker.list_images_in_arv(api_client, 3,
  73                                                                     image_name=image_name,
  74                                                                     image_tag=image_tag)
  75
  76         if not images:
  77             raise WorkflowException("Could not find Docker image %s:%s" % (image_name, image_tag))
  78
  79         pdh = api_client.collections().get(uuid=images[0][0]).execute()["portable_data_hash"]
  80
  81         with cached_lookups_lock:
  82             cached_lookups[dockerRequirement["dockerImageId"]] = pdh
  83
  84     return pdh
  85
  86 def arv_docker_clear_cache():
  87     global cached_lookups
  88     global cached_lookups_lock
  89     with cached_lookups_lock:
  90         cached_lookups = {}