Can someone please post a link to a document or article on installing Cassandra on Docker Swarm?
I think you’ll have to write it
Docker swarm, cassandra cluster, compose
I run 3 Cassandra instances from
docker stack deploy command that reads the following YML file:
version: '3' services: cassandra-1: image: cassandra deploy: restart_policy: condition: on-failure max_attempts: 3 window: 120s placement: constraints: - node.labels.db==cassandra1 environment: CASSANDRA_BROADCAST_ADDRESS: cassandra-1 volumes: - volume1:/var/lib/cassandra ports: - "7000" networks: default: cassandra-2: image: cassandra deploy: restart_policy: condition: on-failure max_attempts: 3 window: 120s placement: constraints: - node.labels.db==cassandra2 environment: CASSANDRA_BROADCAST_ADDRESS: cassandra-2 CASSANDRA_SEEDS: cassandra-1 depends_on: - cassandra-1 volumes: - volume2:/var/lib/cassandra ports: - "7000" networks: default: cassandra-3: image: cassandra deploy: restart_policy: condition: on-failure max_attempts: 3 window: 120s placement: constraints: - node.labels.db==cassandra3 environment: CASSANDRA_BROADCAST_ADDRESS: cassandra-3 CASSANDRA_SEEDS: cassandra-1 depends_on: - cassandra-2 volumes: - volume3:/var/lib/cassandra ports: - "7000" networks: default: volumes: volume1: external: name: cassandra1-vol volume2: external: name: cassandra2-vol volume3: external: name: cassandra3-vol networks: default: external: name: cassandra-net
Alternatively, each instance could be run individually from its
docker service create counterpart with similar parameters.
As a side note, the custom
node.labels.db==cassandra label is required in my setup because of the tight coupling with the volumes in a 3 managers + 5 workers cluster.
For Cassandra, the current docker volume plugin such as folcker, rexray, etc, could not help, as they could not setup the membership among Cassandra members. azaars’s solution is a reasonable way to setup Cassandra on Docker Swarm. While, if the node goes down, one Cassandra member will disappear. You would need additional work to bring up another node, and join the current Cassandra cluster. Cassandra will then recover data automatically.
FireCamp is a recent open source project aims to automate the installation and management of the stateful services, including Cassandra, on Docker Swarm. While, currently it only supports installing the stateful services for Docker Swarm on AWS. Hope it could help your case.
You have created 3 services for Cassandra that means it is fixed to 3 only. what if you want scale up the Cassandra to 4 or 5 ?
$ docker service create --replicas 3 --name cassandra --update-delay 10m cassandra
$ docker service ps cassandra
ID NAME IMAGE NODE DESIRED STATE CURRENT STATE ERROR PORTS
tanytkx5if1g cassandra.1 cassandra:latest pc4 Running Running 5 minutes ago
thg7w2oe25os cassandra.2 cassandra:latest pc2 Running Running 6 minutes ago
i88mbrv12x26 cassandra.3 cassandra:latest pc3 Running Running 5 minutes ago
$ docker service inspect --pretty cassandra
Service Mode: Replicated
On failure: pause
Max failure ratio: 0
Endpoint Mode: vip
network “cassandra-net” is declared as external, but could not be found. You need to create a swarm-scoped network before the stack is deployed
how to create the cassandra-net?