This repository describes the use of custom Docker images for running Apache Cassandra in a cluster of Raspberry Pi's. The cluster management tool used is Docker Swarm.
- Cassandra version: 3.11
- Architecture: armv7l
- The custom Dockerfile in this repository is based on the official Cassandra Dockerfile on the Docker Hub
In a swarm manager node, execute the command line bellow to create a new overlay network named "cassandra-net":
docker network create -d overlay --attachable cassandra-net
Set the type label in nodes to define groups for service scale command:
docker node update --label-add type=(slave or master) node-id
To list with one command all node labels in swarm cluster, you can run:
docker node ls -q | xargs docker node inspect -f '[{{ .Description.Hostname }}]: {{ .Spec.Labels.type }}'
Create the directories that will be mapped as volumes to the Cassandra services data storage.
The default storage path in the host to the Cassandra stack is: /mnt/storage/cassandra. The Docker volume must be map the /var/lib/cassandra directory of the container to the host directory, as follows:
volumes:
- /storage/dir/host:/var/lib/cassandra
To deploy the stack defined in the compose file, execute:
docker stack deploy -c docker-compose.yml cassandra
Two Docker services will be launched, the first being the Cassandra node responsible for the seeds (service seed) and a non-seed node (service node).
Make sure you have clean storage before perform deploy.
To scale the number of Cassandra services, execute the command below, passing the number of replicas (example: 3):
docker service scale cassandra_node=3
As the stack is configured to run only one instance of the Cassandra service (task) per node, it is only possible to scale to a number of replicas equal to the number of available nodes in the cluster.
| Port | Usage |
|---|---|
| 7000 | Unencrypted Intra-node gossip communication |
| 7001 | TLS Intra-node gossip communication |
| 9042 | CQL native service |
| 7199 | JMX server |
| 9160 | Thrift service |
Some configurations are adjustable and provided as environment variables. Default value to some variables are defined in docker-compose.yaml. Those variables are:
| Property Name | Usage |
|---|---|
| CASSANDRA_CLUSTER_NAME | The universal name across all nodes in the cluster |
| CASSANDRA_NUM_TOKENS | Number of tokens for this node. |
| MAX_HEAP_SIZE | Heap size settings (Xms and Xmx) |
| HEAP_NEW_SIZE | Size for young generation (Xmn). |
| SEEDS_SERVICE | Name of the seeds service, used to find the ip of the seed node in the swarm network. |
| TASK_NAME | Name of the task representing each service instance, used to synchronize the boot order |
| WAIT_TIME | Wait time that each task must be wait before starting the cluster join process |
To view the cluster status, among other Cassandra cluster management tasks, use the nodetool utility.
docker exec -t container-id nodetool status
To perform queries in the Cassandra database you can use cqlsh directly, passing the service port and ip of the node that exposes the service:
cqlsh 15.0.0.1 -p9042 --cqlversion=3.4.4
To view Cassandra service runtime log incrementally (like tail -f):
docker service logs -f (service-name or service-id)
- is the same for task-name or task-id