Couchbase XDCR using Docker Swarm, Machine and Compose

Arun GuptaMay 31st, 2016Last Updated: May 29th, 2016

0 20 4 minutes read

Cross Datacenter Replication (XDCR) in Couchbase provides an easy way to replicate data from one cluster to another. The clusters are typically set in geographically diverse data centers. This enables for disaster recovery or to bring data closer to users for faster data access.

This blog will show:

Setup two data centers using Docker Swarm
Run Couchbase containers on each node of Docker Swarm
Setup a Couchbase cluster on each Docker Swarm cluster
Configure one-way XDCR between two Couchbase clusters

For the purpose of this blog, the two data centers will be setup on a local machine using Docker Machine.

Complete code used in this blog is available at: github.com/arun-gupta/couchbase-xdcr-docker.

Create Consul Discovery Service

Each node in Docker Swarm needs to be registered with a discovery service. This blog will use Consul for that purpose. And even Consul will be running on a Docker Machine.

Typically, you’ll run a cluster of Consul but for simplicity a single instance is started in our case.

Create a Docker Machine and start Consul using this script:

 
# Docker Machine for Consul
docker-machine \
   create \
   -d virtualbox \
   consul-machine
 
# Start Consul
docker $(docker-machine config consul-machine) run -d --restart=always \
         -p "8500:8500" \
         -h "consul" \
         progrium/consul -server -bootstrap

Create Docker Swarm cluster

Docker Swarm allows multiple Docker hosts to be viewed as a single unit. This allows your multi-container applications to easily run on multiple hosts. Docker Swarm serves the same Remote API as served by a single host. This allows your existing tools to target a single host or a cluster of hosts.

Both the Docker Swarm clusters will be registered with a single discovery service. This is achieved by using the following value for --swarm-discovery:

consul://$(docker-machine ip consul-machine):8500/v1/kv/<key>

Create a Docker Swarm cluster using Docker Machine using this script:

# Docker Swarm master
docker-machine \
  create \
  -d virtualbox \
  --swarm \
  --swarm-master \
  --swarm-discovery="consul://$(docker-machine ip consul-machine):8500/v1/kv/cluster$1" \
  --engine-opt="cluster-store=consul://$(docker-machine ip consul-machine):8500/v1/kv/cluster$1" \
  --engine-opt="cluster-advertise=eth1:2376" \
  swarm-master-$1
 
# Docker Swarm node-01
docker-machine \
  create \
  -d virtualbox \
  --swarm \
  --swarm-discovery="consul://$(docker-machine ip consul-machine):8500/v1/kv/cluster$1" \
  --engine-opt="cluster-store=consul://$(docker-machine ip consul-machine):8500/v1/kv/cluster$1" \
  --engine-opt="cluster-advertise=eth1:2376" \
  swarm-node-$1-01
 
# Docker Swarm node-02
docker-machine \
  create \
  -d virtualbox \
  --swarm \
  --swarm-discovery="consul://$(docker-machine ip consul-machine):8500/v1/kv/cluster$1" \
  --engine-opt="cluster-store=consul://$(docker-machine ip consul-machine):8500/v1/kv/cluster$1" \
  --engine-opt="cluster-advertise=eth1:2376" \
  swarm-node-$1-02
 
# Configure to use Docker Swarm cluster
eval "$(docker-machine env --swarm swarm-master-$1)"

The script needs to be invoked as:

./create-docker-swarm-cluster.sh A
./create-docker-swarm-cluster.sh B

This will create two Docker Swarm clusters with one “master” and two “worker” as shown below:

NAME              ACTIVE   DRIVER       STATE     URL                         SWARM                     DOCKER    ERRORS
consul-machine    -        virtualbox   Running   tcp://192.168.99.101:2376                             v1.11.1   
default           *        virtualbox   Running   tcp://192.168.99.100:2376                             v1.11.1   
swarm-master-A    -        virtualbox   Running   tcp://192.168.99.102:2376   swarm-master-A (master)   v1.11.1   
swarm-master-B    -        virtualbox   Running   tcp://192.168.99.105:2376   swarm-master-B (master)   v1.11.1   
swarm-node-A-01   -        virtualbox   Running   tcp://192.168.99.103:2376   swarm-master-A            v1.11.1   
swarm-node-A-02   -        virtualbox   Running   tcp://192.168.99.104:2376   swarm-master-A            v1.11.1   
swarm-node-B-01   -        virtualbox   Running   tcp://192.168.99.106:2376   swarm-master-B            v1.11.1   
swarm-node-B-02   -        virtualbox   Running   tcp://192.168.99.107:2376   swarm-master-B            v1.11.1

Consul is running on Docker Machine with IP address 192.168.99.101. And so Consul UI is accessible at http://192.168.99.101:8500:

It shows two Docker Swarm clusters that have been registered.

Exact list of nodes for each cluster can also be seen. Nodes in clusterA are shown:

Nodes in clusterB are shown:

Run Couchbase containers

Run Couchbase container on each node of Docker Swarm cluster using this Compose file.

version: "2"
services:
  db:
    image: arungupta/couchbase
    network_mode: "host"
    ports:
      - 8091:8091
      - 8092:8092
      - 8093:8093
      - 11210:11210

Configure Docker CLI for the first cluster and run 3 containers:

eval "$(docker-machine env --swarm swarm-master-A)"
docker-compose scale db=3

Check the running containers:

> docker ps
CONTAINER ID        IMAGE                 COMMAND                  CREATED             STATUS              PORTS               NAMES
3ec0f15aaee0        arungupta/couchbase   "/entrypoint.sh /opt/"   3 hours ago         Up 3 hours                              swarm-master-A/couchbasexdcrdocker_db_3
07af2ac53539        arungupta/couchbase   "/entrypoint.sh /opt/"   3 hours ago         Up 3 hours                              swarm-node-A-02/couchbasexdcrdocker_db_2
c94878f543fd        arungupta/couchbase   "/entrypoint.sh /opt/"   3 hours ago         Up 3 hours                              swarm-node-A-01/couchbasexdcrdocker_db_1

Configure Docker CLI for the second cluster and run 3 containers:

eval "$(docker-machine env --swarm swarm-master-B)"
docker-compose scale db=3

Check the running containers:

> eval "$(docker-machine env --swarm swarm-master-B)"
> docker ps
CONTAINER ID        IMAGE                 COMMAND                  CREATED             STATUS              PORTS               NAMES
3e3a45480939        arungupta/couchbase   "/entrypoint.sh /opt/"   3 hours ago         Up 3 hours                              swarm-master-B/couchbasexdcrdocker_db_3
1f31f23e337d        arungupta/couchbase   "/entrypoint.sh /opt/"   3 hours ago         Up 3 hours                              swarm-node-B-01/couchbasexdcrdocker_db_1
1feab04c494c        arungupta/couchbase   "/entrypoint.sh /opt/"   3 hours ago         Up 3 hours                              swarm-node-B-02/couchbasexdcrdocker_db_2

Create/Rebalance Couchbase cluster

Scaling and Rebalancing Couchbase Cluster using CLI explains how to create a cluster of Couchbase nodes and rebalance an existing cluster using Couchbase CLI.

Create Couchbase cluster on each Swarm cluster using this script.

export COUCHBASE_CLI=/Users/arungupta/tools/Couchbase-Server-4.0.app/Contents/Resources/couchbase-core/bin/couchbase-cli
for node in 01 02
do
    $COUCHBASE_CLI \
        server-add \
        --cluster=$(docker-machine ip swarm-master-$1):8091 \
        --user Administrator \
        --password password \
        --server-add=$(docker-machine ip swarm-node-$1-$node) \
        --server-add-username=Administrator \
        --server-add-password=password
done
 
$COUCHBASE_CLI \
    setting-cluster \
    --cluster=$(docker-machine ip swarm-master-$1):8091 \
    --user Administrator \
    --password password \
    --cluster-name=cluster$1

The script needs to be invoked as:

./create-couchbase-cluster.sh A

And now rebalance this cluster using this script:

export COUCHBASE_CLI=/Users/arungupta/tools/Couchbase-Server-4.0.app/Contents/Resources/couchbase-core/bin/couchbase-cli
$COUCHBASE_CLI \
rebalance \
--cluster=$(docker-machine ip swarm-master-$1):8091 \
--user Administrator \
--password password \
--server-add-username=Administrator \
--server-add-password=password

This script is invoked as:

./rebalance-couchbase-cluster.sh A

Couchbase Web Console for any node in the cluster will show the output:

Invoke this script to create the second Couchbase cluster as:

./create-couchbase-cluster.sh B

Rebalance this cluster as:

./rebalance-couchbase-cluster.sh B

Couchbase Web Console for any node in the second cluster will show the output:

Setup XDCR

Cross datacenter replication can be setup to be uni-directional, bi-directional or multi-directional. Uni-directional allows data to replicated from source cluster to destination cluster, bi-directional allows replication both ways, multi-directional allows to configure in any direction.

We’ll create a simple uni-directional replication using this script:

export COUCHBASE_CLI=/Users/arungupta/tools/Couchbase-Server-4.0.app/Contents/Resources/couchbase-core/bin/couchbase-cli
$COUCHBASE_CLI \
    xdcr-setup \
    --cluster=$(docker-machine ip swarm-master-$1):8091 \
    --user Administrator \
    --password password \
    --create \
    --xdcr-cluster-name=cluster$1 \
    --xdcr-hostname=$(docker-machine ip swarm-master-$2):8091 \
    --xdcr-username=Administrator \
    --xdcr-password=password \
    --xdcr-demand-encryption=0
 
$COUCHBASE_CLI \
    xdcr-replicate \
    --cluster $(docker-machine ip swarm-master-$1):8091 \
    --xdcr-cluster-name=cluster$1 \
    --user Administrator \
    --password password \
    --create \
    --xdcr-from-bucket=travel-sample \
    --xdcr-to-bucket=travel-sample

This script is invoked as: