marcoabreu commented on a change in pull request #11018: [MXNET-454] Move
distributed Docker cache from S3 to Docker Hub
URL: https://github.com/apache/incubator-mxnet/pull/11018#discussion_r192532129
##########
File path: ci/docker_cache.py
##########
@@ -28,237 +28,164 @@
import logging
import argparse
import sys
-import boto3
-import tempfile
-import pprint
-import threading
-import build as build_util
-import botocore
import subprocess
-from botocore.handlers import disable_signing
-from subprocess import call, check_call, CalledProcessError
+import json
+import build as build_util
from joblib import Parallel, delayed
-S3_METADATA_IMAGE_ID_KEY = 'docker-image-id'
-LOG_PROGRESS_PERCENTAGE_THRESHOLD = 10
-
-cached_aws_session = None
-
-
-class ProgressPercentage(object):
- def __init__(self, object_name, size):
- self._object_name = object_name
- self._size = size
- self._seen_so_far = 0
- self._last_percentage = 0
- self._lock = threading.Lock()
- def __call__(self, bytes_amount) -> None:
- # To simplify we'll assume this is hooked up
- # to a single filename.
- with self._lock:
- self._seen_so_far += bytes_amount
- percentage = int((self._seen_so_far / self._size) * 100)
- if (percentage - self._last_percentage) >=
LOG_PROGRESS_PERCENTAGE_THRESHOLD:
- self._last_percentage = percentage
- logging.info('{}% of {}'.format(percentage, self._object_name))
-
-def build_save_containers(platforms, bucket) -> int:
+def build_save_containers(platforms, registry, load_cache) -> int:
"""
Entry point to build and upload all built dockerimages in parallel
:param platforms: List of platforms
- :param bucket: S3 bucket name
+ :param registry: Docker registry name
+ :param load_cache: Load cache before building
:return: 1 if error occurred, 0 otherwise
"""
if len(platforms) == 0:
return 0
platform_results = Parallel(n_jobs=len(platforms),
backend="multiprocessing")(
- delayed(_build_save_container)(platform, bucket)
+ delayed(_build_save_container)(platform, registry, load_cache)
for platform in platforms)
is_error = False
for platform_result in platform_results:
if platform_result is not None:
- logging.error('Failed to generate {}'.format(platform_result))
+ logging.error('Failed to generate %s', platform_result)
is_error = True
return 1 if is_error else 0
-def _build_save_container(platform, bucket) -> str:
+def _build_save_container(platform, registry, load_cache) -> str:
"""
Build image for passed platform and upload the cache to the specified S3
bucket
:param platform: Platform
- :param bucket: Target s3 bucket
+ :param registry: Docker registry name
+ :param load_cache: Load cache before building
:return: Platform if failed, None otherwise
"""
- docker_tag = build_util.get_docker_tag(platform)
+ docker_tag = build_util.get_docker_tag(platform=platform,
registry=registry)
# Preload cache
- # TODO: Allow to disable this in order to allow clean rebuilds
- load_docker_cache(bucket_name=bucket, docker_tag=docker_tag)
+ if load_cache:
+ load_docker_cache(registry=registry, docker_tag=docker_tag)
# Start building
- logging.debug('Building {} as {}'.format(platform, docker_tag))
+ logging.debug('Building %s as %s', platform, docker_tag)
try:
- image_id = build_util.build_docker(docker_binary='docker',
platform=platform)
- logging.info('Built {} as {}'.format(docker_tag, image_id))
+ image_id = build_util.build_docker(docker_binary='docker',
platform=platform, registry=registry)
+ logging.info('Built %s as %s', docker_tag, image_id)
- # Compile and upload tarfile
- _compile_upload_cache_file(bucket_name=bucket, docker_tag=docker_tag,
image_id=image_id)
+ # Push cache to registry
+ _upload_image(registry=registry, docker_tag=docker_tag,
image_id=image_id)
return None
except Exception:
- logging.exception('Unexpected exception during build of
{}'.format(docker_tag))
+ logging.exception('Unexpected exception during build of %s',
docker_tag)
return platform
# Error handling is done by returning the errorous platform name. This
is necessary due to
# Parallel being unable to handle exceptions
-def _compile_upload_cache_file(bucket_name, docker_tag, image_id) -> None:
+def _upload_image(registry, docker_tag, image_id) -> None:
"""
Upload the passed image by id, tag it with docker tag and upload to S3
bucket
- :param bucket_name: S3 bucket name
+ :param registry: Docker registry name
:param docker_tag: Docker tag
:param image_id: Image id
:return: None
"""
- session = _get_aws_session()
- s3_object = session.resource('s3').Object(bucket_name, docker_tag)
-
- remote_image_id = _get_remote_image_id(s3_object)
- if remote_image_id == image_id:
- logging.info('{} ({}) for {} has not been updated -
skipping'.format(docker_tag, image_id, docker_tag))
- return
- else:
- logging.debug('Cached image {} differs from local {} for
{}'.format(remote_image_id, image_id, docker_tag))
+ _login_dockerhub()
+ # We don't have to retag the image since it is already in the right format
+ logging.info('Uploading %s (%s) to %s', docker_tag, image_id, registry)
+ push_cmd = ['docker', 'push', docker_tag]
+ subprocess.check_call(push_cmd)
- # Compile layers into tarfile
- with tempfile.TemporaryDirectory() as temp_dir:
- tar_file_path = _format_docker_cache_filepath(output_dir=temp_dir,
docker_tag=docker_tag)
- logging.debug('Writing layers of {} to {}'.format(docker_tag,
tar_file_path))
- history_cmd = ['docker', 'history', '-q', docker_tag]
- image_ids_b = subprocess.check_output(history_cmd)
- image_ids_str = image_ids_b.decode('utf-8').strip()
- layer_ids = [id.strip() for id in image_ids_str.split('\n') if id !=
'<missing>']
-
- # docker_tag is important to preserve the image name. Otherwise, the
--cache-from feature will not be able to
- # reference the loaded cache later on. The other layer ids are added
to ensure all intermediary layers
- # are preserved to allow resuming the cache at any point
- cmd = ['docker', 'save', '-o', tar_file_path, docker_tag]
- cmd.extend(layer_ids)
- try:
- check_call(cmd)
- except CalledProcessError as e:
- logging.error('Error during save of {} at {}. Command: {}'.
- format(docker_tag, tar_file_path,
pprint.pprint(cmd)))
- return
-
- # Upload file
- logging.info('Uploading {} to S3'.format(docker_tag))
- with open(tar_file_path, 'rb') as data:
- s3_object.upload_fileobj(
- Fileobj=data,
- Callback=ProgressPercentage(object_name=docker_tag,
size=os.path.getsize(tar_file_path)),
- ExtraArgs={"Metadata": {S3_METADATA_IMAGE_ID_KEY: image_id}})
- logging.info('Uploaded {} to S3'.format(docker_tag))
-
-
-def _get_remote_image_id(s3_object) -> str:
+def _login_dockerhub():
"""
- Get the image id of the docker cache which is represented by the S3 object
- :param s3_object: S3 object
- :return: Image id as string or None if object does not exist
+ Login to the Docker Hub account
+ :return: None
"""
- try:
- if S3_METADATA_IMAGE_ID_KEY in s3_object.metadata:
- cached_image_id = s3_object.metadata[S3_METADATA_IMAGE_ID_KEY]
- return cached_image_id
- else:
- logging.debug('No cached image available for
{}'.format(s3_object.key))
- except botocore.exceptions.ClientError as e:
- if e.response['Error']['Code'] == "404":
- logging.debug('{} does not exist in S3 yet'.format(s3_object.key))
- else:
- raise
+ dockerhub_credentials = _get_dockerhub_credentials()
+ login_cmd = ['docker', 'login', '--username',
dockerhub_credentials['username'], '--password',
+ dockerhub_credentials['password']]
+ subprocess.check_call(login_cmd)
- return None
-
-def load_docker_cache(bucket_name, docker_tag) -> None:
+def load_docker_cache(registry, docker_tag) -> None:
"""
- Load the precompiled docker cache from the passed S3 bucket
- :param bucket_name: S3 bucket name
+ Load the precompiled docker cache from the registry
+ :param registry: Docker registry name
:param docker_tag: Docker tag to load
:return: None
"""
- # Allow anonymous access
- s3_resource = boto3.resource('s3')
- s3_resource.meta.client.meta.events.register('choose-signer.s3.*',
disable_signing)
- s3_object = s3_resource.Object(bucket_name, docker_tag)
-
- # Check if cache is still valid and exists
- remote_image_id = _get_remote_image_id(s3_object)
- if remote_image_id:
- if _docker_layer_exists(remote_image_id):
- logging.info('Local docker cache already present for
{}'.format(docker_tag))
- return
- else:
- logging.info('Local docker cache not present for
{}'.format(docker_tag))
-
- # Download using public S3 endpoint (without requiring credentials)
- with tempfile.TemporaryDirectory() as temp_dir:
- tar_file_path = os.path.join(temp_dir, 'layers.tar')
- s3_object.download_file(
- Filename=tar_file_path,
- Callback=ProgressPercentage(object_name=docker_tag,
size=s3_object.content_length))
-
- # Load layers
- cmd = ['docker', 'load', '-i', tar_file_path]
- try:
- check_call(cmd)
- logging.info('Docker cache for {} loaded
successfully'.format(docker_tag))
- except CalledProcessError as e:
- logging.error('Error during load of docker cache for {} at
{}'.format(docker_tag, tar_file_path))
- logging.exception(e)
- return
- else:
- logging.info('No cached remote image of {} present'.format(docker_tag))
-
-
-def _docker_layer_exists(layer_id) -> bool:
- """
- Check if the docker cache contains the layer with the passed id
- :param layer_id: layer id
- :return: True if exists, False otherwise
- """
- cmd = ['docker', 'images', '-q']
- image_ids_b = subprocess.check_output(cmd)
- image_ids_str = image_ids_b.decode('utf-8').strip()
- return layer_id in [id.strip() for id in image_ids_str.split('\n')]
+ # We don't have to retag the image since it's already in the right format
+ logging.info('Loading Docker cache for %s from %s', docker_tag, registry)
+ pull_cmd = ['docker', 'pull', docker_tag]
+ subprocess.call(pull_cmd) # Don't throw an error if the image does not
exist
-def _get_aws_session() -> boto3.Session: # pragma: no cover
+def delete_local_docker_cache(docker_tag):
"""
- Get the boto3 AWS session
- :return: Session object
+ Delete the local docker cache for the entire docker image chain
+ :param docker_tag: Docker tag
+ :return: None
"""
- global cached_aws_session
- if cached_aws_session:
- return cached_aws_session
-
- session = boto3.Session() # Uses IAM user credentials
- cached_aws_session = session
- return session
+ history_cmd = ['docker', 'history', '-q', docker_tag]
+ try:
+ image_ids_b = subprocess.check_output(history_cmd)
+ image_ids_str = image_ids_b.decode('utf-8').strip()
+ layer_ids = [id.strip() for id in image_ids_str.split('\n') if id !=
'<missing>']
-def _format_docker_cache_filepath(output_dir, docker_tag) -> str:
- return os.path.join(output_dir, docker_tag.replace('/', '_') + '.tar')
+ delete_cmd = ['docker', 'image', 'rm', '--force']
Review comment:
Could you elaborate?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
