375 lines (270 sloc) 13 KB


Notice: The classical python "Docker Registry" is deprecated, in favor of a new golang implementation. This here is kept for historical purpose, and will not receive any significant work/love any more. You should head to the landing page of the new registry or the "Distribution" github project instead.


Build Status

About this document

As the documentation evolves with different registry versions, be sure that before reading any further you:

  • check which version of the registry you are running
  • switch to the corresponding tag to access the README that matches your product version

The stable, released version is the 0.9.1 tag.

Please also have a quick look at the FAQ before reporting bugs.

Table of Contents

Quick start

The fastest way to get running:

That will use the official image from the Docker hub.

Here is a slightly more complex example that launches a registry on port 5000, using an Amazon S3 bucket to store images with a custom path, and enables the search endpoint:

docker run \
         -e SETTINGS_FLAVOR=s3 \
         -e AWS_BUCKET=mybucket \
         -e STORAGE_PATH=/registry \
         -e AWS_KEY=myawskey \
         -e AWS_SECRET=myawssecret \
         -e SEARCH_BACKEND=sqlalchemy \
         -p 5000:5000 \

Configuration mechanism overview

By default, the registry will use the config_sample.yml configuration to start.

Individual configuration options from that file may be overridden using environment variables. Example: docker run -e STORAGE_PATH=/registry.

You may also use different "flavors" from that file (see below).

Finally, you can use your own configuration file (see below).

Configuration flavors

The registry can be instructed to use a specific flavor from a configuration file.

This mechanism lets you define different running "mode" (eg: "development", "production" or anything else).

In the config_sample.yml file, you'll see several sample flavors:

  1. common: used by all other flavors as base settings
  2. local: stores data on the local filesystem
  3. s3: stores data in an AWS S3 bucket
  4. ceph-s3: stores data in a Ceph cluster via a Ceph Object Gateway, using the S3 API
  5. azureblob: stores data in an Microsoft Azure Blob Storage ((docs))
  6. dev: basic configuration using the local flavor
  7. test: used by unit tests
  8. prod: production configuration (basically a synonym for the s3 flavor)
  9. gcs: stores data in Google cloud storage
  10. swift: stores data in OpenStack Swift
  11. glance: stores data in OpenStack Glance, with a fallback to local storage
  12. glance-swift: stores data in OpenStack Glance, with a fallback to Swift
  13. elliptics: stores data in Elliptics key/value storage

You can define your own flavors by adding a new top-level yaml key.

To specify which flavor you want to run, set the SETTINGS_FLAVOR environment variable: export SETTINGS_FLAVOR=dev

The default flavor is dev.

NOTE: it's possible to load environment variables from within the config file with a simple syntax: _env:VARIABLENAME[:DEFAULT]. Check this syntax in action in the example below...

Example config

common: &common
    standalone: true
    loglevel: info
    search_backend: "_env:SEARCH_BACKEND:"

    <<: *common
    loglevel: warn
    storage: s3
    s3_access_key: _env:AWS_S3_ACCESS_KEY
    s3_secret_key: _env:AWS_S3_SECRET_KEY
    s3_bucket: _env:AWS_S3_BUCKET
    boto_bucket: _env:AWS_S3_BUCKET
    storage_path: /srv/docker
    smtp_host: localhost

    <<: *common
    loglevel: debug
    storage: local
    storage_path: /home/myself/docker

    <<: *common
    storage: local
    storage_path: /tmp/tmpdockertmp

Available configuration options

When using the config_sample.yml, you can pass all options through as environment variables. See config_sample.yml for the mapping.

General options

  1. loglevel: string, level of debugging. Any of python's logging module levels: debug, info, warn, error or critical
  2. debug: boolean, make the /_ping endpoint output more useful information, such as library versions and host information.
  3. storage_redirect: Redirect resource requested if storage engine supports this, e.g. S3 will redirect signed URLs, this can be used to offload the server.
  4. boto_host/boto_port: If you are using storage: s3 the standard boto config file locations (/etc/boto.cfg, ~/.boto) will be used. If you are using a non-Amazon S3-compliant object store (such as Ceph), in one of the boto config files' [Credentials] section, set boto_host, boto_port as appropriate for the service you are using. Alternatively, set boto_host and boto_port in the config file.

Authentication options

  1. standalone: boolean, run the server in stand-alone mode. This means that the Index service on will not be used for anything. This implies disable_token_auth.

  2. index_endpoint: string, configures the hostname of the Index endpoint. This is used to verify passwords of users that log in. It defaults to You should probably leave this to its default.

  3. disable_token_auth: boolean, disable checking of tokens with the Docker index. You should provide your own method of authentication (such as Basic auth).

Search-engine options

The Docker Registry can optionally index repository information in a database for the GET /v1/search endpoint. You can configure the backend with a configuration like:

The search_backend setting selects the search backend to use. If search_backend is empty, no index is built, and the search endpoint always returns empty results.

  1. search_backend: The name of the search backend engine to use. Currently supported backends are:
    1. sqlalchemy

If search_backend is neither empty nor one of the supported backends, it should point to a module.


  search_backend: foo.registry.index.xapian

In this case, the module is imported, and an instance of its Index class is used as the search backend.


Use SQLAlchemy as the search backend.

  1. sqlalchemy_index_database: The database URL passed through to create_engine.


  search_backend: sqlalchemy
  sqlalchemy_index_database: sqlite:////tmp/docker-registry.db

On initialization, the SQLAlchemyIndex class checks the database version. If the database doesn't exist yet (or does exist, but lacks a version table), the SQLAlchemyIndex creates the database and required tables.

Mirroring Options

All mirror options are placed in a mirroring section.

  1. mirroring:
    1. source:
    2. source_index:
    3. tags_cache_ttl:


    tags_cache_ttl: 172800 # 2 days

Beware that mirroring only works for the public registry. You can not create a mirror for a private registry.

Cache options

It's possible to add an LRU cache to access small files. In this case you need to spawn a redis-server configured in LRU mode. The config file "config_sample.yml" shows an example to enable the LRU cache using the config directive cache_lru.

Once this feature is enabled, all small files (tags, meta-data) will be cached in Redis. When using a remote storage backend (like Amazon S3), it will speed things up dramatically since it will reduce roundtrips to S3.

All config settings are placed in a cache or cache_lru section.

  1. cache/cache_lru:
    1. host: Host address of server
    2. port: Port server listens on
    3. password: Authentication password

Storage options

storage selects the storage engine to use. The registry ships with two storage engine by default (file and s3).

If you want to find other (community provided) storages: pip search docker-registry-driver

To use and install one of these alternate storages:

  • pip install docker-registry-driver-NAME
  • in the configuration set storage to NAME
  • add any other storage dependent configuration option to the conf file
  • review the storage specific documentation for additional dependency or configuration instructions.

    Currently, we are aware of the following storage drivers:

storage file

  1. storage_path: Path on the filesystem where to store data


  storage: file
  storage_path: /mnt/registry

Persistent storage

If you use any type of local store along with a registry running within a docker remember to use a data volume for the storage_path. Please read the documentation for data volumes for more information.


docker run -p 5000 -v /tmp/registry:/tmp/registry registry

storage s3

AWS Simple Storage Service options

  1. s3_access_key: string, S3 access key
  2. s3_secret_key: string, S3 secret key
  3. s3_bucket: string, S3 bucket name
  4. s3_region: S3 region where the bucket is located
  5. s3_encrypt: boolean, if true, the container will be encrypted on the server-side by S3 and will be stored in an encrypted form while at rest in S3.
  6. s3_secure: boolean, true for HTTPS to S3
  7. s3_use_sigv4: boolean, true for USE_SIGV4 (boto_host needs to be set or use_sigv4 will be ignored by boto.)
  8. boto_bucket: string, the bucket name for non-Amazon S3-compliant object store
  9. boto_host: string, host for non-Amazon S3-compliant object store
  10. boto_port: for non-Amazon S3-compliant object store
  11. boto_debug: for non-Amazon S3-compliant object store
  12. boto_calling_format: string, the fully qualified class name of the boto calling format to use when accessing S3 or a non-Amazon S3-compliant object store
  13. storage_path: string, the sub "folder" where image data will be stored.


  storage: s3
  s3_region: us-west-1
  s3_bucket: acme-docker
  storage_path: /registry
  s3_access_key: AKIAHSHB43HS3J92MXZ
  s3_secret_key: xdDowwlK7TJajV1Y7EoOZrmuPEJlHYcNP2k4j49T

Your own config

Start from a copy of config_sample.yml.

Then, start your registry with a mount point to expose your new configuration inside the container (-v /home/me/myfolder:/registry-conf), and point to it using the DOCKER_REGISTRY_CONFIG environment variable:

sudo docker run -p 5000:5000 -v /home/me/myfolder:/registry-conf -e DOCKER_REGISTRY_CONFIG=/registry-conf/mysuperconfig.yml registry

Advanced use

For more features and advanced options, have a look at the advanced features documentation


For more backend drivers, please read

For developers

Read contributing