Cluster Reconfiguration

Create Service Config Files

Use the following command to create a service specific configuration yaml.

accelo admin makeconfig <service_config_name>

Use accelo info to get respective service config name, except for following services:

Update Pulse Configs

Application Config

All cluster configuration information configure during accelo config cluster is stored under file config/acceldata<cluster_name>.conf

Host Config

All the host level configuration details are stored in vars.yml file and the host list is stored in hydra_hosts.yml file.

Accelo Reconfig Cluster

This command regenerates the pulse configuration files with the updated information from the cluster.

The list of files which are generated are:

Overriding Default vars.yml

In order to overwrite existing configurations present in vars.yml create a file named as override.yml under work/<clustername>

Vars.yml

base:

    enable_ntp_stats: "false"

    ntp_daemon: ntpd

Override.yml

base:

  enable_ntp_stats: "true"

  ntp_daemon: "chronyd"

Following command overrides vars.yml to required property values.

#here -a is optional for running across all configured clusters

accelo reconfig cluster -a

$ accelo reconfig cluster -h

Dynamic Reconfiguration of the Cluster

​

Usage:override.yml

  accelo reconfig cluster [flags]

  accelo reconfig cluster [command]

​

Available Commands:

  vars        Archive the vars.yml and hostRoleMap

​

Flags:

  -a, --all    All clusters to be reconfigured

  -h, --help   help for cluster

  -s, --skip   Skip the reconfiguring of the cluster and will just update the Epoch time and reload the hydra server

For reconfiguring cluster run the following commands:

accelo reconfig cluster -a

Any changes in the acceldata config file requires config push to DB using the following command:

accelo admin database push-config

Agents, Base, Static, Components, and JMX

To override the Agent Section, add the agent section to the override.yml file and specify the fields with the values you wish to change.

For example: Suppose you want the pulsenodeagent to have a 60-second interval. The override.yml file will look like the following:

agent:

  interval: 60s

Similarly, you can override the Base, Hydra, Static, Components, and JMX sections.

Extra Plugins

Extra Plugins has three individual subsections:

To override a subsection of the Extra Plugins Section, specify the subsection you want to change with the modified values.

For example: In the jmxagent subsection, addhive server interactive appended to the previous values (HBase master, HBase regionserver, data node, namenode, journalnode).

extra_plugins:

  jmxagent:

    - hive_server_interactive

    - hbase_master

    - hbase_regionserver

    - datanode

    - namenode

    - journalnode

EnabledJMX

These two sections contain a list of plugins that are enabled during the config cluster process.

1. extra_plugins
2. enabled_jmx

To uninstall a single or several plugins, add the section to the override.yml file.

For example: If you want to add hive_metastore in the enabled_jmx section do the following:

enabled_jmx:

  - hive_metastore

If you want to remove the hive_metastore from the enabled_ jmx section copy the section from the vars.yml file and paste it in the override.yml file and remove the hive_metastore line and run the accelo reconfig cluster.

enabled_jmx:

 - hive_server

 - zookeeper_server

 - zeppelin_master

 - kafka_broker

Hosts

The Host to the installed component Role map is found in the Hosts section.

If you wish to add a host that is not already in the Ambari, Cloudera, or standalone cluster, you can do so in the override.yml file.

For example:

hosts:

  host1:

  - nifi_server

LogLocation

Add the log title to the enabled logs section if you wish to add a LogLocation that was not previously enabled. Add the subsection from the vars.yml file with that change.

For example: If you want to enable the Kafka_err logs then the override.yml file will look like the following block.

enabled_logs:

 - kafka_err

 - hive_server

 - hive_metastore

log_locations:

   kafka_err:

    - path: /var/log/kafka/kafka.err

      type: KAFKA_ERR

On removing the kafka_err logs the override.yml file will look like this:

enabled_logs:

 - hive_server

 - hive_metastore

Elastic connection timeout from ad-graphql

A new configuration named ELASTIC_CONNECTION_TIMEOUT has been introduced, with a default value set to 60 seconds. This configuration allows control over the connection timeout from ad-graphql to Elasticsearch.

Add the variable ELASTIC_CONNECTION_TIMEOUT to the env section of ad-graphql in the $AcceloHome/config/docker/ad-core.yml file. After adding the variable, your ad-core.yml file should appear as follows:

ad-graphql:

    image: ad-graphql

    container_name: ""

    environment:

    - ELASTIC_CONNECTION_TIMEOUT=10000