Infinispan's new image

Infinispan 10 introduced a new server, which does not utilise the same launch commands and configuration as the legacy 9.4 wildfly based server. Therefore, we decided that this was an excellent opportunity to rewrite our container image from scratch to better suite the capabilities of the new server and to provide all the functionality required by the Infinispan Operator.

This post focuses on the server image’s architecture. Future blog posts will focus on more advanced configurations, as well as example usage and deployment scenarios such as deploying a cluster using Kubernetes StatefulSets.

Show me the code!

The source code for the Infinispan image has a new home. The image can be found at Currently this repository only contains the server image, however our intention is for this to also be the home for all future Infinispan related images.

Where’s the Dockerfile?

The most noticable change when looking at the repository is that there is no Dockerfile in the source tree. This is because we decided to utilise the open-source tool CEKit to build our images. CEKit is an image creation tool that allows container images to be created using multiple build engines (e.g. docker, Buildah, Podman) with a single configuration. Installation instructions can be found here, but the basic command to create a Docker based image is as follows.

cekit build docker

CEKit leverages .yaml files for all configuration, opposed to a Dockerfile, as this allows for build time overriding of image properties. For example, with CEKit it’s possible to override the server artifact version without modifying any files, instead the following is passed as a build parameter.

cekit build --overrides '{"artifacts": [{"name": "", "path": ""}]}' docker

More detailed instructions about how to build the server image from source can be found in the image’s documentation.

Ok so where can I get a pre-built image?

Previously the Infinispan images were deployed exclusively under the jboss namespace at jboss/infinispan-server, however this repository has now been deprecated and will be removed eventually.

Instead, all Infinispan images will now be released under the infinispan namespace and are hosted at both and Docker Hub, as and infinispan/server.

Getting Started

To get started with infinispan server on your local machine simply execute:

docker run -p 11222:11222 infinispan/server

By default the image has authentication enabled on all exposed endpoints. When executing the above command the image automatically generates a username/password combo, prints the values to stdout and then starts the Infinispan server with the authenticated Hotrod and Rest endpoints exposed on port 11222. Therefore, it’s necessary to utilise the printed credentials when attempting to access the exposed endpoints via clients.

It’s also possible to provide a username/password combination via environment variables like so:

docker run -p 11222:11222 -e USER="Titus Bramble" -e PASS="Shambles" infinispan/server

Connecting via Hotrod

Using the credentials passed in the command above, it is now possible to connect via the HotRod client using the following file. Note, that the following SASL properties must be configured on your client, with the username and password properties changed as required, otherwise the connection will fail:

infinispan.remote.auth-username=Titus Bramble

Connecting via REST

The REST endpoint is configured to utilise the DIGEST protocol, therefore it’s necessary for the HTTP client to authenticate requests. For example, the name of all caches can be retrieved via the following curl command:

 curl --digest -u "Titus Bramble:Shambles" http://localhost:11222/rest/v2/cache

Further Reading

For more detailed information abou how to use the image, please consult the official documentation.

In the next blog post we will look at how the server can be configured for more advanced use-cases by supplying configuration and identity .yaml files.

Get it, Use it, Ask us!

Please download, report bugs, chat with us, ask questions on StackOverflow.

Posted by Ryan Emerson on 2019-12-02
Tags: docker
back to top