Public | Automated Build

Last pushed: a year ago
Short Description
Dockerhub image sync tool for cloud (openstack, opennebula) catalogs.
Full Description


This project is a continuation of the discontinued


This component can be used to synchronize Docker images in DockerHub to instances of either OpenStack or OpenNebula with Docker support already configured. To do so, it relies on the WebHooks mechanisms offered by DockerHub infrastructure and it provides a REST interface to list the available images already synchronized and some operations to force the synchronization of individual ones.


This project uses the following elements:

The later one is not present in any maven repository and so it's included in the project to easen the deployment.

Additionally the REST interface is based on Jersey 2.X and Grizzly

All dependencies save OpenNebula java bindings are managed by maven automatically


The repository synchronization project needs access to:

  • One docker installation, prefereably in the same machine as it's running but it might be in any remote location
  • One OpenStack or OpenNebula installation which is accessible from the host in which the project is running


The configuration files should be defined in /etc/indigo-reposync directory however, if the server or client must be run as a user instead of root, they can be stored in $HOME/.indigo-reposync folder.
This is the folder's content:

The configuration properties which are accepted are described in

This file is the main configuration one and should contain the following mandatory properties:

  • REPOSYNC_TOKEN: The secret token used for authorization which will be used in the DockerHub webhook. It will be used in the rest of the operations as an authorization token as well.
  • REPOSYNC_REST_ENDPOINT: Endpoint for the REST server to listen for requests
  • REPOSYNC_BACKEND: Backend to use, which should be one value of OpenNebula or OpenStack

Depending on the backend, further properties will have to be defined.

OpenStack configuration

When using OpenStack as a target for the synchronization, the following properties should be defined in file:

OpenNebula configuration

When using OpenNebula as backend, the following properties should be defined in the configuration file:

  • ONE_XMLRPC: URL pointing to the RPC endpoint of the OpenNebula installation
  • ONE_AUTH: Path to an authorization file in the filesystem containing the OpenNebula credentials in the format <username>:<password>

This file is a standard JRE logging configuration file that will be used to log events in the REST service.


A file containing a list of repositories to synchronize when the sync operation is executed. It must contain a repository per line.


To start developing just clone the repository with git clone and create a valid configuration file. Compile the code with mvn compile

To execute the REST server, run ./ start


When using the development environment, packages for Ubuntu 14.04/16.04 and CentOS 7 can be generated by using the deb and rpm profiles respectively. Notice that to successfuly build a RPM package, you'll need to be running a Linux distribution that allows to install the rpm utility.
Once this packages are built, they can be installed and they will provide a binary named indigo-reposync that can be run both as server, executing indigo-reposync start and as client executing any other command. They will provide a default configuration as follow:

  • Using a local Docker installation that must be accessible through /var/run/docker.sock socket
  • Logging to /var/log/reposync0.log file
  • No content for repolist file
  • The configuration in file MUST be completed with a valid OpenNebula or OpenStack access or the server won't start

This configuration can be overriden or customized to fill any other needs

Docker image

To easen deployment, a docker file is provided. To create a docker image with the reposync project, execute the following command in the root of java-syncrepos project:

docker build -t indigo-reposync:v1 .

It will create an image named indigo-reposync with tag v1. Please, feel free to choose whatever name and tag you may want.

To run it, access to the host docker installation is needed. To do so, please run the recently created image with the command:

docker run -v /var/run/docker.sock:/var/run/docker.sock --name reposync -i -t indigo-reposync:v1

Some stub configuration files are included in the image and they can be filled and used to run the REST service, however, it's recommended to manage them in the host system and mount them as volumes. Also, to be able to access the service from
outside the host system, port redirection might be needed. A common startup configuration can be:

docker run -d -v /var/run/docker.sock:/var/run/docker.sock -v /etc/indigo-reposync:/etc/indigo-reposync -v /<homedir>/.one/one_auth:/root/one_auth -p 80:8085 --name reposyncserver -i -t indigo-reposync:v1

That way, the configuration is passed from the host system to the container and it can be shared among different instances.

To execute any command, when the container is running execute:

docker exec reposyncserver indigo-reposync <command>

The following commands are available in the reposync shell script:

  • start: Starts the REST server
  • list: List the images in the backend IaaS platform
  • pull <image> [tag]: Pull an image with an optional tag from DockerHub into the IaaS repository.
  • delete <imageId>: Delete an image from the IaaS repository
  • sync: Execute a synchronization operation pulling all images and tags found in the repository list specified in the REPOSYNC_REPO_LIST_FILE file

DockerHub configuration

In order to benefit from the automatic synchronization of DockerHub repositories, some webhooks are needed that will trigger the pulling of the updated images.
Let's imagine that we run the docker image with the following command:

docker run -d -v /var/run/docker.sock:/var/run/docker.sock -v /etc/indigo-reposync:/etc/indigo-reposync -v /home/ubuntu/lipcaroot.pem:/root/lipcaroot.pem -p 80:8085 --name reposyncserver -i -t indigodatacloud/reposync:latest

This will run the server taking the configuration from the directory /home/ubuntu/.indigo-reposync in the host system and let's assume that this configuration makes it listen on port 8085 and that the host system has port 80 open to the outside world. Also, let's assume that the host machine has a DNS name '' and that we've defined a secret token in the configuration file as 'mysecrettoken'.

If we wanted to have the images in DockerHub repository indigodatacloud/ubuntu-sshd synchronized then we would need to create it pointing to the following URL:

Now everytime that an image is pushed to the indigodatacloud/ubuntu-sshd repository, the reposync service will receive a notification and execute a pull operation over the image tags that have been updated, integrating and/or updating them in the local image repository (in this case, Glance from OpenStack).
If the reposync service is down when the notification is launched or the network is not working, then a manual pull can be executed by accessing the machine in which the reposync docker container is running and executing docker exec reposyncserver reposync pull indigodatacloud/ubuntu-sshd <updated_tag> where \<updated_tag> is the name of the tag that has just been pushed (and whose notification has been missed).

Docker Pull Command
Source Repository