- Overview
- Setup - The basics of getting started with Cassandra
- Usage - Configuration options and additional functionality
- Reference
- Limitations - OS compatibility, etc.
- Development
A Puppet module to install and manage Cassandra, DataStax Agent & OpsCenter
- Installs the Cassandra package (default cassandra22 on Red Hat and cassandra on Debian).
- Configures settings in
${config_path}/cassandra.yaml. - On CentOS 7 if the
initservice provider is used, then cassandra is added as a system service. - Optionally ensures that the Cassandra service is enabled and running.
- On Debian systems:
- Optionally replace
/etc/init.d/cassandrawith a workaround for CASSANDRA-9822.
- Optionally replace
- Optionally installs the DataStax agent.
- Optionally sets JAVA_HOME in /etc/default/datastax-agent.
- Optionally configures a Yum repository to install the Cassandra packages from (on Red Hat).
- Optionally configures an Apt repository to install the Cassandra packages from (on Debian).
- Optionally configures files in the
/etc/dsedirectory if one is using DataStax Enterprise.
- Optionally configures the firewall for the Cassandra related network ports.
- Optionally installs a JRE/JDK package (e.g. java-1.7.0-openjdk) and the Java Native Access (JNA).
- Optionally installs the Cassandra support tools (e.g. cassandra22-tools).
We follow SemVer Versioning and an update of the major release (i.e. from 1.Y.Z to 2.Y.Z) will indicate a significant change to the API which will most probably require a change to your manifest.
This is a major change to the API and you will more than likely need to change your manifest to accomodate these changes.
The service_ensure attribute of the cassandra class now defaults to
undef, users who do want to manage service status in Puppet can still set
it to true. If leaving the value at the default and setting
service_refresh and service_enable to false will mean that the
user and not Puppet running will control the running state of
Cassandra. This currently works OK on the Red Hat family, but
has issues on Debian due to
CASSANDRA-2356
during an initial install or package upgrade.
All the functionality relating to OpsCenter has been divested to the locp/opscenter module on Puppet Forge.
It should also be noted that the module no longer creates directories for
the data, commitlog, saved_caches and for Cassandra 3 the hints
directory. These resources will now need to be defined in your
manifest/profile.
For a list of features that have been deprecated in this release, please see https://github.com/voxpupuli/puppet-cassandra/wiki/Deprecations
For details on migrating from the version 1.X.X attributes to the settings
hash, see
(https://github.com/voxpupuli/puppet-cassandra/wiki/Suggested-Baseline-Settings)
Please also see the notes for 2.0.0 in the CHANGELOG.
The hints_directory documentation will cause a change in the cassandra.yaml file regardless of the value you set it to. If you do not wish this to result in a refesh of the Cassandra service, please set service_refresh to false.
Now that Cassandra 3 is available from the DataStax repositories, there is a problem (especially on Debian) with the operating system package manager attempting to install Cassandra 3. This can be mitigated against using something similar to the code in this modules acceptance test. Please note that the default Cassandra package name has now been changed from 'dsc'. See the documentation for cassandra::package_name below for details.
if $::osfamily == 'RedHat' {
$version = '2.2.4-1'
} else {
$version = '2.2.4'
}
class { 'cassandra':
package_ensure => $version,
}A somewhat embarrassing correction to the spelling of the cassandra::fail_on_non_suppoted_os to cassandra::fail_on_non_supported_os.
Unfortunately both releases 1.3.7 and 1.4.0 have subsequently been found to call a refresh service even when no changes had been made to the underlying configuration. In release 1.8.0 (somewhat belatedly) the service_refresh flag has been introduced to mitigate against similar problems.
- Please see the notes for 1.4.0.
- cassandra::cassandra_package_ensure has been renamed to cassandra::package_ensure.
- cassandra::cassandra_package_name has been renamed to cassandra::package_name.
There is now a cassandra::datastax_agent class, therefore:
- cassandra::datastax_agent_package_ensure has now been replaced with cassandra::datastax_agent::package_ensure.
- cassandra::datastax_agent_service_enable has now been replaced with cassandra::datastax_agent::service_enable.
- cassandra::datastax_agent_service_ensure has now been replaced with cassandra::datastax_agent::service_ensure.
- cassandra::datastax_agent_package_name has now been replaced with cassandra::datastax_agent::package_name.
- cassandra::datastax_agent_service_name has now been replaced with cassandra::datastax_agent::service_name.
Likewise now there is a new class for handling the installation of Java:
- cassandra::java_package_ensure has now been replaced with cassandra::java::ensure.
- cassandra::java_package_name has now been replaced with cassandra::java::package_name.
Also there is now a class for installing the optional utilities:
- cassandra::cassandra_opt_package_ensure has now been replaced with cassandra::optutils:ensure.
- cassandra::cassandra_opt_package_name has now been replaced with cassandra::optutils:package_name.
-
cassandra_opt_package_ensure changed from 'present' to undef.
-
The manage_service option has been replaced with service_enable and service_ensure.
Create a Cassandra 2.X cluster called MyCassandraCluster which uses the GossipingPropertyFileSnitch and password authentication. In this very basic example the node itself becomes a seed for the cluster and the credentials will default to a user called cassandra with a password called of cassandra.
# Cassandra pre-requisites
include cassandra::datastax_repo
include cassandra::java
class { 'cassandra':
settings => {
'authenticator' => 'PasswordAuthenticator',
'cluster_name' => 'MyCassandraCluster',
'commitlog_directory' => '/var/lib/cassandra/commitlog',
'commitlog_sync' => 'periodic',
'commitlog_sync_period_in_ms' => 10000,
'data_file_directories' => ['/var/lib/cassandra/data'],
'endpoint_snitch' => 'GossipingPropertyFileSnitch',
'listen_address' => $::ipaddress,
'partitioner' => 'org.apache.cassandra.dht.Murmur3Partitioner',
'saved_caches_directory' => '/var/lib/cassandra/saved_caches',
'seed_provider' => [
{
'class_name' => 'org.apache.cassandra.locator.SimpleSeedProvider',
'parameters' => [
{
'seeds' => $::ipaddress,
},
],
},
],
'start_native_transport' => true,
},
require => Class['cassandra::datastax_repo', 'cassandra::java'],
}However, PLEASE note that this is the ABSOLUTE MINIMUM configuration to get Cassandra up and running but will probably give you a rather badly configured node. Please see Suggested Baseline Settings for details on making your configuration a lot more robust.
For this code to run with version 3.X of Cassandra, the hints_directory will
also need to be specified:
...
class { 'cassandra':
settings => {
...
'hints_directory' => '/var/lib/cassandra/hints',
...
},
require => Class['cassandra::datastax_repo', 'cassandra::java'],
}In your top level node classification (usually common.yaml), add the
settings hash and all the tweaks you want all the clusters to use:
cassandra::baseline_settings:
authenticator: AllowAllAuthenticator
authorizer: AllowAllAuthorizer
auto_bootstrap: true
auto_snapshot: true
...Then, in the individual node classification add the parts which define the cluster:
cassandra::settings:
cluster_name: developer playground cassandra cluster
cassandra::dc: Onsite1
cassandra::rack: RAC1
cassandra::package_ensure: 3.0.5-1
cassandra::package_name: cassandra30We assume that authentication has been enabled for the cassandra cluster and we are connecting with the default user name and password ('cassandra/cassandra').
In this example, we create a keyspace (mykeyspace) with a table called 'users' and an index called 'users_lname_idx'.
We also add three users (to Cassandra, not the mykeyspace.users table) called spillman, akers and boone while ensuring that a user called lucan is absent.
class { 'cassandra':
...
}
class { 'cassandra::schema':
cqlsh_password => 'cassandra',
cqlsh_user => 'cassandra',
cqlsh_host => $::ipaddress,
indexes => {
'users_lname_idx' => {
table => 'users',
keys => 'lname',
keyspace => 'mykeyspace',
},
},
keyspaces => {
'mykeyspace' => {
durable_writes => false,
replication_map => {
keyspace_class => 'SimpleStrategy',
replication_factor => 1,
},
}
},
permissions => {
'Grant select permissions to spillman to all keyspaces' => {
permission_name => 'SELECT',
user_name => 'spillman',
},
'Grant modify to to keyspace mykeyspace to akers' => {
keyspace_name => 'mykeyspace',
permission_name => 'MODIFY',
user_name => 'akers',
},
'Grant alter permissions to mykeyspace to boone' => {
keyspace_name => 'mykeyspace',
permission_name => 'ALTER',
user_name => 'boone',
},
'Grant ALL permissions to mykeyspace.users to gbennet' => {
keyspace_name => 'mykeyspace',
permission_name => 'ALTER',
table_name => 'users',
user_name => 'gbennet',
},
},
tables => {
'users' => {
columns => {
user_id => 'int',
fname => 'text',
lname => 'text',
'PRIMARY KEY' => '(user_id)',
},
keyspace => 'mykeyspace',
},
},
users => {
'spillman' => {
password => 'Niner27',
},
'akers' => {
password => 'Niner2',
superuser => true,
},
'boone' => {
password => 'Niner75',
},
'gbennet' => {
'password' => 'foobar',
},
'lucan' => {
'ensure' => absent
},
},
}In the DataStax documentation Initializing a multiple node cluster (single data center) http://docs.datastax.com/en/cassandra/2.2/cassandra/initialize/initSingleDS.html there is a basic example of a six node cluster with two seeds to be created in a single data center spanning two racks. The nodes in the cluster are:
| Node Name | IP Address |
|---|---|
| node0 (seed 1) | 110.82.155.0 |
| node1 | 110.82.155.1 |
| node2 | 110.82.155.2 |
| node3 (seed 2) | 110.82.156.3 |
| node4 | 110.82.156.4 |
| node5 | 110.82.156.5 |
Each node is configured to use the GossipingPropertyFileSnitch and 256 virtual nodes (vnodes). The name of the cluster is MyCassandraCluster. Also, while building the initial cluster, we are setting the auto_bootstrap to false.
In this initial example, we are going to expand the example by:
- Ensuring that the software is installed via the DataStax Community
repository by including
cassandra::datastax_repo. This needs to be executed before the Cassandra package is installed. - That a suitable Java Runtime environment (JRE) is installed with Java Native
Access (JNA) by including
cassandra::java. This need to be executed before the Cassandra service is started.
node /^node\d+$/ {
class { 'cassandra::datastax_repo':
before => Class['cassandra']
}
class { 'cassandra::java':
before => Class['cassandra']
}
class { 'cassandra':
settings => {
'authenticator' => 'AllowAllAuthenticator',
'auto_bootstrap' => false,
'cluster_name' => 'MyCassandraCluster',
'commitlog_directory' => '/var/lib/cassandra/commitlog',
'commitlog_sync' => 'periodic',
'commitlog_sync_period_in_ms' => 10000,
'data_file_directories' => ['/var/lib/cassandra/data'],
'endpoint_snitch' => 'GossipingPropertyFileSnitch',
'hints_directory' => '/var/lib/cassandra/hints',
'listen_interface' => 'eth1',
'num_tokens' => 256,
'partitioner' => 'org.apache.cassandra.dht.Murmur3Partitioner',
'saved_caches_directory' => '/var/lib/cassandra/saved_caches',
'seed_provider' => [
{
'class_name' => 'org.apache.cassandra.locator.SimpleSeedProvider',
'parameters' => [
{
'seeds' => '110.82.155.0,110.82.156.3',
},
],
},
],
'start_native_transport' => true,
},
}
}The default value for the num_tokens is already 256, but it is included in the example for clarity. Do not forget to either set auto_bootstrap to true or not set the attribute at all after initializing the cluster.
To continue with the examples provided by DataStax, we look at the example for a cluster across multiple data centers http://docs.datastax.com/en/cassandra/2.2/cassandra/initialize/initMultipleDS.html.
| Node Name | IP Address | Data Center | Rack |
|---|---|---|---|
| node0 (seed 1) | 10.168.66.41 | DC1 | RAC1 |
| node1 | 10.176.43.66 | DC1 | RAC1 |
| node2 | 10.168.247.41 | DC1 | RAC1 |
| node3 (seed 2) | 10.176.170.59 | DC2 | RAC1 |
| node4 | 10.169.61.170 | DC2 | RAC1 |
| node5 | 10.169.30.138 | DC2 | RAC1 |
For the sake of simplicity, we will confine this example to the nodes:
node /^node[012]$/ {
class { 'cassandra':
dc => 'DC1',
settings => {
'authenticator' => 'AllowAllAuthenticator',
'auto_bootstrap' => false,
'cluster_name' => 'MyCassandraCluster',
'commitlog_directory' => '/var/lib/cassandra/commitlog',
'commitlog_sync' => 'periodic',
'commitlog_sync_period_in_ms' => 10000,
'data_file_directories' => ['/var/lib/cassandra/data'],
'endpoint_snitch' => 'GossipingPropertyFileSnitch',
'hints_directory' => '/var/lib/cassandra/hints',
'listen_interface' => 'eth1',
'num_tokens' => 256,
'partitioner' => 'org.apache.cassandra.dht.Murmur3Partitioner',
'saved_caches_directory' => '/var/lib/cassandra/saved_caches',
'seed_provider' => [
{
'class_name' => 'org.apache.cassandra.locator.SimpleSeedProvider',
'parameters' => [
{
'seeds' => '110.82.155.0,110.82.156.3',
},
],
},
],
'start_native_transport' => true,
},
}
}
node /^node[345]$/ {
class { 'cassandra':
dc => 'DC2',
settings => {
'authenticator' => 'AllowAllAuthenticator',
'auto_bootstrap' => false,
'cluster_name' => 'MyCassandraCluster',
'commitlog_directory' => '/var/lib/cassandra/commitlog',
'commitlog_sync' => 'periodic',
'commitlog_sync_period_in_ms' => 10000,
'data_file_directories' => ['/var/lib/cassandra/data'],
'endpoint_snitch' => 'GossipingPropertyFileSnitch',
'hints_directory' => '/var/lib/cassandra/hints',
'listen_interface' => 'eth1',
'num_tokens' => 256,
'partitioner' => 'org.apache.cassandra.dht.Murmur3Partitioner',
'saved_caches_directory' => '/var/lib/cassandra/saved_caches',
'seed_provider' => [
{
'class_name' => 'org.apache.cassandra.locator.SimpleSeedProvider',
'parameters' => [
{
'seeds' => '110.82.155.0,110.82.156.3',
},
],
},
],
'start_native_transport' => true,
},
}
}We don't need to specify the rack name (with the rack attribute) as RAC1 is the default value. Again, do not forget to either set auto_bootstrap to true or not set the attribute at all after initializing the cluster.
After configuring the relevant repository, the following snippet works on CentOS 7 to install DSE Cassandra 4.7.0, set the HADOOP_LOG_DIR, set the DSE_HOME and configure DataStax Enterprise to use LDAP for authentication:
class { 'cassandra::datastax_repo':
descr => 'DataStax Repo for DataStax Enterprise',
pkg_url => 'https://username:[email protected]/enterprise',
before => Class['cassandra'],
}
class { 'cassandra':
cluster_name => 'MyCassandraCluster',
config_path => '/etc/dse/cassandra',
package_ensure => '4.7.0-1',
package_name => 'dse-full',
service_name => 'dse',
...
}
class { 'cassandra::dse':
file_lines => {
'Set HADOOP_LOG_DIR directory' => {
ensure => present,
path => '/etc/dse/dse-env.sh',
line => 'export HADOOP_LOG_DIR=/var/log/hadoop',
match => '^# export HADOOP_LOG_DIR=<log_dir>',
},
'Set DSE_HOME' => {
ensure => present,
path => '/etc/dse/dse-env.sh',
line => 'export DSE_HOME=/usr/share/dse',
match => '^#export DSE_HOME',
},
},
settings => {
ldap_options => {
server_host => localhost,
server_port => 389,
search_dn => 'cn=Admin',
search_password => secret,
use_ssl => false,
use_tls => false,
truststore_type => jks,
user_search_base => 'ou=users,dc=example,dc=com',
user_search_filter => '(uid={0})',
credentials_validity_in_ms => 0,
connection_pool => {
max_active => 8,
max_idle => 8,
}
}
}
}DataStax announced in late October 2016 that it was no longer supporting
the community edition of Cassandra or DSC as it was known (see
[Take a bow Planet
Cassandra](http://www.datastax.com/2016/10/take-a-bow-planet-cassandra)
for details). However, the following snippet of code running on Ubuntu
14.04 worked fine without having to change any of the ::cassandra class
settings:
require cassandra::java
include cassandra::optutils
class { 'cassandra::apache_repo':
release => '310x',
before => Class['cassandra', 'cassandra::optutils'],
}
class { 'cassandra':
...
}The reference documentation is generated using the puppet-strings tool. To see all of it, please go to http://voxpupuli.github.io/puppet-cassandra.
- When using a Ruby version before 1.9.0, the contents of the Cassandra configuration file may change order of elements due to a problem with to_yaml in earlier versions of Ruby.
- When creating key spaces, indexes, cql_types and users the settings will only be used to create a new resource if it does not currently exist. If a change is made to the Puppet manifest but the resource already exits, this change will not be reflected.
- At the moment the
cassandra::system::transparent_hugepagedoes not persist between reboots. - Acceptance for Debian 7 are confined to Cassandra 2.1 and 2.2. There is a conflict between the GLIBC on Debian 7 and the newer releases of Cassandra 3.X.
Contributions will be gratefully accepted. Please go to the project page, fork the project, make your changes locally and then raise a pull request. Details on how to do this are available at https://guides.github.com/activities/contributing-to-open-source.
Please also see the CONTRIBUTING.md page for project specific requirements.
For a list of contributers see CONTRIBUTING.md and https://github.com/voxpupuli/puppet-cassandra/graphs/contributors