Charmed Kafka Documentation - How to manage units

charm
, , ,
1

How to manage units

Unit management guide for scaling and running admin utility scripts.

Replication and Scaling

Increasing the number of Kafka brokers can be achieved by adding more units to the Charmed Kafka application, for example:

juju add-unit kafka -n <num_brokers_to_add>

For more information on how to manage units, please refer to the Juju documentation

It is important to note that when adding more units, the Kafka cluster will not automatically rebalance existing topics and partitions. New storage and new brokers will be used only when new topics and new partitions are created.

Partition reassignment can still be done manually by the admin user by using the charmed-kafka.reassign-partitions Kafka bin utility script. Please refer to its documentation for more information.

IMPORTANT Scaling down is currently not supported in the charm automation.
If partition reassignment is not manually performed before scaling down in order to make sure the decommissioned units do not hold any data, your cluster may suffer to data loss.

Running Kafka admin utility scripts

Apache Kafka ships with bin/*.sh commands to do various administrative tasks such as:

  • bin/kafka-config.sh to update cluster configuration
  • bin/kafka-topics.sh for topic management
  • bin/kafka-acls.sh for management of ACLs of Kafka users

Please refer to the upstream Kafka project, for a full list of the bash commands available in Kafka distributions. Also, you can use --help argument for printing a short summary of the argument for a given bash command.

The most important commands are also exposed via the Charmed Kafka snap, accessible via charmed-kafka.<command>. Please refer to this table for more information about the mapping between the Kafka bin commands and the snap entrypoints.

IMPORTANT Before running bash scripts, make sure that some listeners have been correctly opened by creating appropriate integrations. Please refer to this table for more information about how listeners are opened based on relations. To simply open a SASL/SCRAM listener, just integrate a client application using the data-integrator, as described here.

To run most of the scripts, you need to provide:

  1. the Kafka service endpoints, generally referred to as bootstrap servers
  2. authentication information

Juju admins of the Kafka deployment

For Juju admins of the Kafka deployment, the bootstrap servers information can be obtained using

BOOTSTRAP_SERVERS=$(juju run kafka/leader get-admin-credentials | grep "bootstrap.servers" | cut -d "=" -f 2)

Admin client authentication information is stored in the /var/snap/charmed-kafka/common/etc/kafka/client.properties file present on every Kafka broker. The content of the file can be accessed using

juju ssh kafka/leader `cat /etc/kafka/client.properties`

This file can be provided to the Kafka bin commands via the --command-config argument. Note that client.properties may also refer to other files ( e.g. truststore and keystore for TLS-enabled connections). Those files also need to be accessible and correctly specified.

Commands can also be run within a Kafka broker, since both the authentication file (along with the truststore if needed) and the Charmed Kafka snap are already present.

Example (listing topics)

For instance, in order to list the current topics on the Kafka cluster, you can run:

juju ssh kafka/leader 'charmed-kafka.topics --bootstrap-server $BOOTSTRAP_SERVERS --list --command-config /var/snap/charmed-kafka/common/etc/kafka/client.properties'

Juju external users

For external users managed by the Data Integrator Charm, the endpoints and credentials can be fetched using the dedicated action

juju run data-integrator/leader get-credentials --format yaml

The client.properties file can be generated by substituting the relevant information in the file available on the brokers at /var/snap/charmed-kafka/current/etc/kafka/client.properties

To do so, fetch the information using juju commands:

BOOTSTRAP_SERVERS=$(juju run data-integrator/leader get-credentials --format yaml | yq .kafka.endpoints )
USERNAME=$(juju run data-integrator/leader get-credentials --format yaml | yq .kafka.username )
PASSWORD=$(juju run data-integrator/leader get-credentials --format yaml | yq .kafka.password )

Then copy the /var/snap/charmed-kafka/current/etc/kafka/client.properties and substitute the following lines:

...
sasl.jaas.config=org.apache.kafka.common.security.scram.ScramLoginModule required username="<USERNAME>" password="<PASSWORD>";
...
bootstrap.servers=<BOOTSTRAP_SERVERS>