Couchbase Cluster with docker and docker-compose

Hint: on Linux use sudo before docker.

Configure

Cluster accepts the following parameters:

• ADMIN_LOGIN
• ADMIN_PASSWORD (must be minimum 8 chars with special chars)
• CLUSTER_RAM_QUOTA (recommended minimum 2048)
• INDEX_RAM_QUOTA (recommended minimum 256)
• MODEL_BUCKET_RAMSIZE (recommended minimum 256)
• FILE_BUCKET_RAMSIZE (recommended minimum 256)

For what the couchbase memory parameters mean, see Couchbase Cluster Settings documentation

Run

1. git clone https://github.com/etops/couchbase-cluster-setup.git
2. create ~/couchbase/node1 folders
3. docker-compose up -d (remove the -d to see log output)
4. go to http://<ip>:8091 and verify running couchbase

<ip> can be found with docker inspect couchbaseclustersetup_couchbase-init_1 | grep IPAddress | cut -d '"' -f 4

Scale

docker-compose scale couchbase=3 Syntaxt docker-compose scale SERVICE=#Nr

Total number of running instance is #Nr + 1

Data location

use volume in docker-compose.yml

volumes:
        - ~/couchbase/node1:/opt/couchbase/var

Problem: docker-compose scale and volumes don't work together Workaround:

couchbase1:
    build: .
    links:
        - couchbase-init:couchbase
    environment: 
        - CLUSTER_RAM_QUOTA=1024
    restart: always
    volumes:
        - ~/couchbase/node1:/opt/couchbase/var
    
couchbase2:
    build: .
    links:
        - couchbase-init:couchbase
    environment: 
        - CLUSTER_RAM_QUOTA=1024
    restart: always
    volumes:
        - ~/couchbase/node2:/opt/couchbase/var
    
couchbase3:
    build: .
    links:
        - couchbase-init:couchbase
    environment: 
        - CLUSTER_RAM_QUOTA=1024
    restart: always
    volumes:
        - ~/couchbase/node3:/opt/couchbase/var

Change Password

Password can be set in Dockerfile as environement variable.

Stop and rebuild

stop all running containers: docker-compose stop

remove images of the stoped containers: docker-compose rm

force build: docker-compose build

run again: docker-compose up

Cluster Design Considerations

Couchbase guidance suggests keeping number of total buckets to a minimum, ideally 5 or less, never more than 10.

They recommend starting in one bucket and growing out as necessary.

example docker swarm

version: "3.3"

services:
  couchbase-master:
    environment:
      - TYPE=MASTER
      - CLUSTER_RAM_QUOTA=5120
      - ADMIN_LOGIN=admin
      - ADMIN_PASSWORD=<pw>
    image: etops/couchbase:4.6.2-rawdata
    ports:
      - target: 4369
        published: 4369
        protocol: tcp
        mode: host
      - target: 8091
        published: 8091
        protocol: tcp
        mode: host
      - target: 8092
        published: 8092
        protocol: tcp
        mode: host
      - target: 8093
        published: 8093
        protocol: tcp
        mode: host
      - target: 11210
        published: 11210
        protocol: tcp
        mode: host
    volumes:
      - type: bind
        source: /home/ubuntu/stack/rawdata/couchbase/data
        target: /opt/couchbase/var
    networks:
      - cbrd
    deploy:
      mode: replicated
      replicas: 1
      placement:
        constraints: [node.labels.app_role == rawdata]
      restart_policy:
        condition: on-failure

  couchbase-worker:
    environment:
      - TYPE=WORKER
      - COUCHBASE_MASTER=159.100.242.154
      - ADMIN_LOGIN=admin
      - ADMIN_PASSWORD=<pw>
    image: etops/couchbase:4.6.2-rawdata
    ports:
      - target: 4369
        published: 4369
        protocol: tcp
        mode: host
      - target: 8091
        published: 8091
        protocol: tcp
        mode: host
      - target: 8092
        published: 8092
        protocol: tcp
        mode: host
      - target: 8093
        published: 8093
        protocol: tcp
        mode: host
      - target: 11210
        published: 11210
        protocol: tcp
        mode: host
    volumes:
      - type: bind
        source: /home/ubuntu/stack/rawdata/couchbase/data
        target: /opt/couchbase/var
    networks:
      - cbrd
    deploy:
      mode: replicated
      replicas: 1
      placement:
        constraints: [node.labels.app_role == rawdata]
      restart_policy:
        condition: on-failure

networks:
  cbrd:

Minimum Requirements

Couchbase's published guidance can be found here.

Deployment Best Practices

See notes here.