cassandra

ansible-role-cassandra

Ansible role to install and configure Apache Cassandra.

Requirements

OS Family Identification

The Ansible playbook must gather enough facts to have the ansible_os_family variable to be set.

Python Modules

On Debian systems that set cassandra_configure_apache_repo to True please note that the apt_repository module is used. Therefore the below requirements are needed on the host that executes this module.

python-apt (python 2)
python3-apt (python 3)

Role Variables

cassandra_15770_workaround: (default: False) CASSANDRA-15770 affected Debian 10 and Ubuntu 20.04 it was subsequently fixed in versions 3.0.21, 3.11.7, 4.0. As we are running our tests against 3.11.6 at the moment, this applies the fix to /etc/init.d/cassandra if set to True.
cassandra_2356_workaround: (default: False) CASSANDRA-2356 affected deb packages for Cassandra versions pre-4.0.
cassandra_cms_heap_new_size_mb: A custom fact that returns a value (MB) that might be suitable to set the HEAP_NEWSIZE when using the Concurrent Mark Sweep (CMS) Collector. See Tuning Java resources for more details. Requires the cassandra_memtotal_mb and cassandra_processor_vcpus facts to be set.
cassandra_cms_max_heapsize_mb: A custom fact that returns a value (MB) that might be suitable to set the MAX_HEAP_SIZE when using the Concurrent Mark Sweep (CMS) Collector. See Tuning Java resources for more details. Requires the cassandra_memtotal_mb fact to be set.
cassandra_configuration (default: none): The configuration for Cassandra. See the example play book below.
cassandra_configuration_directory (default: /etc/cassandra/default.conf on RedHat and /etc/cassandra in Debian): The directory of the Cassandra configuration.
cassandra_configuration_file (default: /etc/cassandra/default.conf/cassandra.yaml on RedHat and /etc/cassandra/cassandra.yaml on Debian): The location of the Cassandra configuration file.
cassandra_configuration_templ (default: cassandra.yaml.j2): Name of the template file for Cassandra configuration.
cassandra_configure_apache_repo (default: False): Whether to configure the Apache Cassandra repository.

SEE ALSO: cassandra_repo_apache_release.
cassandra_dc: If defined will set the Datacenter in cassandra-rackdc.properties.

This option is deprecated and will be removed in a future release. Please use cassandra_regex_replacements instead.
cassandra_directories: If defined, this will create directories after the package has been installed (therefore when the cassandra user is available) but before Cassandra is configured. See the example playbook for more details.

Please note that when used with the cassandra_join_cluster option the paths specified here will be deleted (with all their contents) and recreated so that the node can join a cluster correctly.
cassandra_heap_new_size_mb: A custom fact that returns a value (MB) that might be suitable to set the HEAP_NEWSIZE. See Tuning Java resources for more details. Requires the cassandra_memtotal_mb and cassandra_processor_vcpus facts to be set.
cassandra_install_packages (default: True): A boolean value indicating if this Ansible role should attempt to install packages or not. If set to False it allows a user to use this role to configure Cassandra but does not attempt to install it (e.g. they have installed from source code).
cassandra_join_cluster (default: False): This is an experimental feature, use with extreme caution! If set to true and the cassandra_node_count is found to be 1 on a node, the following actions will take place on that node before the node is configured:
- Cassandra will be stopped
- The Cassandra data directory/directories will be cleared of all data.
When that has been done, the node will be reconfigured and join a cluster (which should be larger in size than one node).
cassandra_max_heapsize_mb: A custom fact that returns a value (MB) that might be suitable to set the MAX_HEAP_SIZE. See Tuning Java resources for more details. Requires the cassandra_memtotal_mb fact to be set.
cassandra_memtotal_mb (default: ansible_memtotal_mb if set): Is used to calculate cassandra_cms_max_heapsize_mb, cassandra_max_heapsize_mb, cassandra_cms_heap_new_size_mb and cassandra_heap_new_size_mb.
cassandra_path This variable was introduced in version 1.4.0 but was missed from being documented. As pointed out if the variable is defined when the variable is set and the directory is deleted, it is not recreated. The variable has now been deprecated in favour of cassandra_directories.
cassandra_node_count: A read-only variable that attempts to contain the number of nodes in the cluster. Will only be set if cassandra_join_cluster is set to True.
cassandra_package (default: cassandra): The name of the package to be installed to provide Cassandra.
cassandra_processor_vcpus (default: ansible_processor_vcpus if set): Is used to calculate cassandra_cms_heap_new_size_mb and cassandra_heap_new_size_mb.
cassandra_rack: If defined will set the rack in cassandra-rackdc.properties.

This option is deprecated and will be removed in a future release. Please use cassandra_regex_replacements instead.
cassandra_regex_replacements (default: []): A list of hashes describing a path which is relative to cassandra_configuration_directory, regexp which is a regular expression to find in a file and line is the replacement within the file. See the example playbook for more details.
cassandra_repo_apache_release (default: None): The name of the release series (can be one of 40x, 311x, or 30x). This must be set if cassandra_configure_apache_repo is set to True.
cassandra_service_enabled (default: yes): Should the cassandra service be enabled (can be yes or no).
cassandra_service_restart (default: True): If set to true, changes to the Cassandra configuration file or the data directories will ensure that Cassandra service is refreshed after the changes. Setting this flag to false will disable this behaviour, therefore allowing the changes to be made but allow the user to control when the service is restarted.
cassandra_systemd_enabled (default: False): Should the Cassandra service by enabled via systemd if set to True then a unit file will be placed in cassandra_systemd_path created from template cassandra_systemd_template.
cassandra_systemd_path (default: /usr/lib/systemd/system/cassandra.service): The path to a unit file for Cassandra. This variable is ignored if cassandra_systemd_enabled is False.
cassandra_systemd_template (default: "systemd/system/cassandra.service.j2": The path for a template from which to create unit file for Cassandra. This variable is ignored if cassandra_systemd_enabled is False.
cassandra_task_delay (default: 10): Ansible tasks that rely on network connectivity (i.e. apt, apt_key, apt_repository, package and yum) may be affected by high latency so those tasks are configured to retry. This parameter is how long to wait between attempts. See also cassandra_task_retries.
cassandra_task_retries (default: 5): Ansible tasks that rely on network connectivity (i.e. apt, apt_key, apt_repository, package and yum) may be affected by high latency so those tasks are configured to retry. This parameter is how many attempts should be made. See also cassandra_task_delay.

Example Playbook

This playbook should be enough to configure Cassandra with a very basic configuration:

---
- name: Example Playbook for the locp.cassandra Role

  hosts: cassandra

  remote_user: root

  vars:
    cassandra_configuration:
      authenticator: PasswordAuthenticator
      cluster_name: MyCassandraCluster
      commitlog_directory: /data/cassandra/commitlog
      commitlog_sync: periodic
      commitlog_sync_period_in_ms: 10000
      data_file_directories:
        - /data/cassandra/data
      endpoint_snitch: GossipingPropertyFileSnitch
      hints_directory: "/data/cassandra/hints"
      listen_address: "{{ ansible_default_ipv4.address }}"
      # For a sensible value to set num_tokens to, please see
      # https://cassandra.apache.org/doc/latest/getting_started/production.html#tokens
      num_tokens: 4
      partitioner: org.apache.cassandra.dht.Murmur3Partitioner
      saved_caches_directory: /data/cassandra/saved_caches
      seed_provider:
        - class_name: "org.apache.cassandra.locator.SimpleSeedProvider"
          parameters:
            # This sets the seed to the first node in the Ansible group called
            # cassandra.
            - seeds: "{{ hostvars[groups['cassandra'][0]]['ansible_default_ipv4'].address }}"
      start_native_transport: true
    cassandra_configure_apache_repo: true
    # Create an alternative directories structure for the Cassandra data.
    # In this example, the will be a directory called /data owned by root
    # with rwxr-xr-x permissions.  It will have a series of sub-directories
    # all of which will be defaulted to being owned by the cassandra user
    # with rwx------ permissions.
    cassandra_directories:
      root:
        group: root
        mode: "0755"
        owner: root
        paths:
          - /data
      data:
        paths:
          - /data/cassandra
          - /data/cassandra/commitlog
          - /data/cassandra/data
          - /data/cassandra/hints
          - /data/cassandra/saved_caches
    cassandra_regex_replacements:
      - path: cassandra-env.sh
        line: 'MAX_HEAP_SIZE="{{ cassandra_max_heapsize_mb }}M"'
        regexp: '^#MAX_HEAP_SIZE="4G"'
      - path: cassandra-env.sh
        line: 'HEAP_NEWSIZE="{{ cassandra_heap_new_size_mb }}M"'
        regexp: '^#HEAP_NEWSIZE="800M"'
      - path: cassandra-rackdc.properties
        line: 'dc=DC1'
        regexp: '^dc='
      - path: cassandra-rackdc.properties
        line: 'rack=RACK1'
        regexp: '^rack='
    cassandra_repo_apache_release: 40x

  roles:
    - role: locp.cassandra

To see the playbooks that are used for testing see molecule/default/converge.yml and molecule/combine_cluster/converge.yml.