infinispan

Install Infinispan or Red Hat DataGrid server configurations.

Requirements

Infinispan 16.1+ requires Java 25, which is only available on RHEL 9.7+ and RHEL 10.1+ For more details please go through the official Infinispan 16.1 release notes

RHEL 8 is not supported with Infinispan 16.1+ (Java 25 is not available), More details
For RHEL 8 systems, use Infinispan 16.0.x or earlier versions

Role Defaults

Service configuration defaults

Variable	Description	Default
`infinispan_port`	Alternate port for the service	`11222`
`infinispan_jgroups_port`	Alternate port for the jgroups cluster	`7800`
`infinispan_jgroups_relay_port`	Alternate port for the jgroups relaying cluster	`7801`
`infinispan_port_offset`	Optional port offset for colocated installations	`0`
`infinispan_nodename`	Instance name for service (ie. cluster node identifier)	`{{ inventory_hostname }}`
`infinispan_keycloak_persistence`	Enable persitence datasource for keycloak caches	`false`
`infinispan_service_user`	Posix account for the service installation	`ispn`
`infinispan_service_group`	Posix group for the service installation	`ispn`
`infinispan_logfile_format`	Main logfile format: FILE or JSON-FILE	`FILE`
`infinispan_logfile_root_level`	Root logging level: TRACE, DEBUG, INFO, WARN, ERROR	`INFO`
`infinispan_logfile_enable_audit`	Enable additional audit.log logfile	`true`
`infinispan_logfile_enable_hotrod_accesslog`	Enable additional hotrod-access.log	`false`
`infinispan_logfile_enable_rest_accesslog`	Enable additional rest-access.log	`false`
`infinispan_logfile_maxsize`	Max file size, triggers rotation	`100 MB`
`infinispan_default_realm_tls`	Enable TLS server certificate	`false`
`infinispan_keystore_path`	Path to keystore containing server identity certificate	`/etc/pki/java/cacerts`
`infinispan_keystore_password`	Keystore password	`changeit`
`infinispan_keystore_alias`	Alias for the certificate to load from keystore	`{{ inventory_hostname }}`
`infinispan_keystore_key_password`	Key passphrase for TLS server identity	`''`
`infinispan_keycloak_caches`	Creates remote caches for keycloak	`false`
`infinispan_jvm_package`	RHEL java package runtime	`java-25-openjdk-headless`
`infinispan_service_name`	Name of the systemd service unit, appended with `-{{infinispan_port_offset}}` when not 0	`infinispan`
`infinispan_service_desc`	Description of the systemd service unit	`Infinispan`
`infinispan_service_restart_on_failure`	systemd restart-on-failure behavior activation	`true``
`infinispan_service_startlimitintervalsec`	systemd StartLimitIntervalSec	`300` if `infinispan_service_restart_on_failure` else ``
`infinispan_service_startlimitburst`	systemd StartLimitBurst	`5` if `infinispan_service_restart_on_failure` else ``
`infinispan_service_restartsec`	systemd RestartSec	`10s` if `infinispan_service_restart_on_failure` else ``
`infinispan_resp_cache`	Name of the cache on which to enable the RESP protocol; if empty, disable RESP	`''`

Cluster configuration

Variable	Description	Default
`infinispan_jgroups_relay`	Enable cross-DC relaying	`false`
`infinispan_jgroups_relay_sites`	List of site names for cross-DC relaying	`[]`
`infinispan_jgroups_relay_site`	Site the inventory host is in when cross-DC is enabled	`''`
`infinispan_jgroups_discovery`	Clustering discovery protocol, value from [`PING`,`TCPPING`,`JDBC_PING`]	``
`infinispan_jgroups_iface`	The NIC name to be used for cluster IPv4 addresses (ie. ‘eth0’)	`default_ipv4`
`infinispan_jgroups_cluster_nodes`	List of node definitions for jgroups cluster, read below for the format	`[]`

The infinispan_jgroups_cluster_nodes parameter, when empty, tell the collection to geenrate the list from the hosts variables in ansible_play_hosts; otherwise, it can be passed-in using the following dictionary format:

infinispan_jgroups_cluster_nodes:
  - address: 10.0.0.175
    inventory_host: '10.0.0.175[7800]'
    name: us-east-2-datagrid-1
    port: 7800
    site: us-east-2
    value: 'tcp://10.0.0.175:7800'
  - address: 10.0.0.179
    inventory_host: "10.0.0.179[7800]"
    name: us-east-2-datagrid-2
    port: 7800
    site: us-east-2
    value: 'tcp://10.0.0.179:7800'

where address, port, and inventory_host are connection details; name is the name of the host in the cluster, site is the name of the cluster in the xsite configuration, and value is explicit connection string.

Download and install defaults

Variable	Description	Default
`infinispan_offline_install`	Perform an offline install	`false`
`infinispan_version`	Infinispan version to install	`16.1.4`
`infinispan_bundle`	Archive name for Infinispan download	`infinispan-server-{{ infinispan_version }}.zip`
`infinispan_download_url`	Download URL for infinispan	`https://github.com/infinispan/infinispan/releases/download/{{ infinispan_version }}/{{ infinispan_bundle }}`
`infinispan_dest`	Directory where to extract installation archives	`/opt/infinispan`
`infinispan_installation_path`	Specific unxtracted installation path for infinispan	`/opt/infinispan/infinispan-server-{{ infinispan_version }}/`
`infinispan_app_download_dir`	Directory where to download archives	`/opt/infinispan`
`infinispan_healthcheck`	Check health of service at end of installation	`true`
`infinispan_bind_address`	Alternate bind address for the daemon	`localhost`
`infinispan_caches`	List of cache definitions to configure statically	`[]`
`infinispan_users`	List of users to create	`[]`
`infinispan_rest_cache_api_path`	Path of infinispan rest api	`/rest/v2/caches/`
`infinispan_configure_firewalld`	Ensure firewalld is running and configure infinispan ports	`false`

Role Variables

The following are a set of required variables for the role:

Variable	Description	Required
`infinispan_supervisor_password`	Password for the administration console user account	`true`
`infinispan_users`	List of user definitions to create	`false`
`infinispan_java_home`	JAVA_HOME of installed JRE, leave empty for using specified `infinispan_jvm_package` RPM path	`false`

Sample user definition format:

infinispan_users:
  - { name: 'testuser1', password: 'test', roles: 'observer' }
  - { name: 'testuser2', password: 'test', roles: 'application' }

The following are required when infinispan_jgroups_discovery is JDBC_PING:

Variable	Description	Default
`infinispan_jdbc_engine`	backend database engine (values: `['mariadb','postgres','sqlserver']`)	`mariadb`
`infinispan_jdbc_driver_version`	driver version to download	`2.7.4`
`infinispan_jdbc_url`	URL for jdbc connection	`jdbc:mariadb://localhost:3306/keycloak`
`infinispan_jdbc_user`	username for jdbc connection	`keycloak-user`
`infinispan_jdbc_pass`	password for jdbc connection	`keycloak-pass`

When setting up cross-DC relaying, also setup mariadb in active-active mode (ie. with galera cluster), and switch the JDBC to url to the sequential scheme; similar configuration for other database engines.

Dependencies

The role depends on the following collections:

To install, from the collection root directory, run:

ansible-galaxy collections install -r requirements.yml

Python lxml and jmespath libraries are needed on the host that executes this module. To install, from the collection root directory, run:

pip install -r requirements.txt

Example Playbook

The following is an example playbook that makes use of the role to install Infinispan

---
- hosts: ...
      collections:
        - middleware_automation.infinispan
      tasks:
        - name: Include Infinispan role
          include_role:
            name: infinispan
          vars:
            infinispan_supervisor_password: "changeme"
            infinispan_users: []

Offline installation

Performing an offline installation is possible by:

setting infinispan_offline_install to true
making the file available to ansible controller, using as filename infinispan_bundle.

Keycloak integration

Enabling infinispan_keycloak_caches will prepare the following caches in a dedicated cache container for remote keycloak access:

- sessions
- offlineSessions
- clientSessions
- offlineClientSessions
- loginFailures
- actionTokens
- work

for more details, refer to the: INSTALLATION AND CONFIGURATION GUIDE - 3.4.6. Infinispan caches

Deploying custom cache configurations

It is possible to deploy caches statically in the infinispan.xml server configuration, by populating the infinispan_caches list of dicts. In case of clustered deployments, this configuration must be executed against all nodes in the cluster, as they are not propagated by the service. To deploy cache configurations at runtime, refer to the infinispan_cache role instead.

For more details, refer to:

How to configure caches

A cache configuration can be passed-in as an XML file or string, as it is generated by the infinispan server console; otherwise, it is possible to define configuration dicts, used to populate the available templates.

To pass XML directly, as string or with a file:

infinispan_caches:
  - cache_xml: >
          <local-cache name="testcachexml" statistics="true">
            <encoding media-type="application/x-protostream"/>
          </local-cache>
  - cache_xml: "{{ lookup('file', user_data_file) }}"

To configure caches:

infinispan_caches:
  - cache_config:
      name: 'my-distributed-cache'
      template: 'distribute-cache'
      encoding: 'application/x-jboss-marshalling'
      persistence: false

It is possible to mix and match the two configurations in the infinispan_caches list.

Vector Search Support

Infinispan 15+ natively supports vector search for AI/ML use cases, allowing k-Nearest Neighbor (kNN) queries on vector embeddings.

To enable vector search, your client application must:

Define ProtoStream entities with @Vector annotations specifying dimension and similarity algorithm
Register the ProtoStream schema with Infinispan when connecting
Create indexed caches (either via the client API or using the templates in this role with indexing: true and indexing_entities pointing to your registered schemas)

The Ansible role supports configuring indexed caches through the indexing parameter:

infinispan_cache_config:
  name: 'vector-search-cache'
  template: 'distributed'
  encoding: 'application/x-protostream'
  indexing: true
  indexing_storage: 'filesystem'
  indexing_entities:
    - 'com.example.YourEntity'  # Must be registered first by your application

Note: Infinispan requires at least one indexed entity when indexing is enabled. Your application must register the ProtoStream schema before creating the indexed cache, or the cache creation will fail.