These instructions assume being executed locally on a Linux or macOS machine using the docker-machine. Actual separate servers can be utilised as well, just use their IPs when selecting the manager node and joining workers. Also please make sure the latest docker is installed on all those nodes.
Create a manager node:
$ docker-machine create manager1
Create two more worker nodes:
$ docker-machine create worker1
$ docker-machine create worker2
Make sure all 3 nodes are up and running:
$ docker-machine ls
NAME ACTIVE DRIVER STATE URL SWARM DOCKER ERRORS
manager1 - virtualbox Running tcp://192.168.99.100:2376 v18.02.0-ce
worker1 - virtualbox Running tcp://192.168.99.101:2376 v18.02.0-ce
worker2 - virtualbox Running tcp://192.168.99.102:2376 v18.02.0-ce
Get manager1 IP address, we'll use it to promote the manager node and join the rest to it:
$ docker-machine ip manager1
192.168.99.100
SSH to the manager machine and init the swarm:
$ docker-machine ssh manager1
$ docker swarm init --advertise-addr 192.168.99.100
Swarm initialized: current node (00seko4p1wfide9hgx0hhllfg) is now a manager.
To add a worker to this swarm, run the following command:
docker swarm join --token SWMTKN-1-25iz6yloiemiadfnig8staiw3llcx45odbnwb3k4spoo94s9ft-4j7no435z27mc32733q8tdqbm 192.168.99.100:2377
To add a manager to this swarm, run 'docker swarm join-token manager' and follow the instructions.
$ docker node ls
ID HOSTNAME STATUS AVAILABILITY MANAGER STATUS
00seko4p1wfide9hgx0hhllfg * manager1 Ready Active Leader
Make sure to note the token value.
Now let's join worker nodes to the manager (make sure to replace the token and IP with your actual values):
$ docker-machine ssh worker1
$ docker swarm join --token SWMTKN-1-25iz6yloiemiadfnig8staiw3llcx45odbnwb3k4spoo94s9ft-4j7no435z27mc327
33q8tdqbm 192.168.99.100:2377
This node joined a swarm as a worker.
$ docker-machine ssh worker2
$ docker swarm join --token SWMTKN-1-25iz6yloiemiadfnig8staiw3llcx45odbnwb3k4spoo94s9ft-4j7no435z27mc327
33q8tdqbm 192.168.99.100:2377
This node joined a swarm as a worker.
Let's check that the swarm is configured correctly, from the manager node:
$ docker node ls
ID HOSTNAME STATUS AVAILABILITY MANAGER STATUS
00seko4p1wfide9hgx0hhllfg * manager1 Ready Active Leader
ofs056wa4y31wtl9fp5rhcswn worker1 Ready Active
6az80xjz05nyhboz59mcl43pc worker2 Ready Active
From now on, everything will be done on the manager node.
On the manager node, clone the TykTechnologies/tyk-pro-docker-demo:
$ git clone https://github.com/TykTechnologies/tyk-pro-docker-demo.git
Open the tyk_analytics.conf file in the confs/ folder and add your license string to the "license_key": "" section. Edit any other configurations if needed. By default it's ready for a simple deployment.
Note that Docker Swarm does not work with multiple compose files that build on each other, instead all the desired compose files must be merged in one. We already provide a default in docker-swarm-merged.yml, however when doing changes to the base or swarm-specific or your own custom compose files, they need to be merged manually using the docker-compose program:
$ docker-compose -f docker-compose.yml -f docker-swarm.yml config > docker-swarm-merged.yml
The resulting file is the one to be used with the docker stack commands. Note that it may contain absolute paths.
Once ready, deploy the stack to the swarm:
$ docker stack deploy --compose-file docker-swarm-merged.yml tyk-pro-demo
Note that you can update your stack using the same command, unless changes to configs are made because docker treats them as immutable objects. In this case, either the stack needs to be removed and deployed again, or a configuration object needs to be renamed in the docker-compose file.
Eventually, all the containers should be launched and ready:
$ docker service ls
ID NAME MODE REPLICAS IMAGE PORTS
omzcwm2r01s0 tyk-pro-demo_tyk-dashboard replicated 2/2 tykio/tyk-dashboard:latest *:3000->3000/tcp, *:5000->5000/tcp
poctpcc8xr6t tyk-pro-demo_tyk-gateway replicated 2/2 tykio/tyk-gateway:latest *:8080->8080/tcp
4p8k5wxn8fgy tyk-pro-demo_tyk-mongo replicated 1/1 mongo:3.2 *:27017->27017/tcp
vfoh5zjez2fd tyk-pro-demo_tyk-pump replicated 2/2 tykio/tyk-pump-docker-pub:latest
8bvfjijoqzy6 tyk-pro-demo_tyk-redis replicated 1/1 redis:latest *:6379->6379/tcp
$ docker stack ps tyk-pro-demo
ID NAME IMAGE NODE DESIRED STATE CURRENT STATE ERROR PORTS
jnzmgf9n88ug tyk-pro-demo_tyk-pump.1 tykio/tyk-pump-docker-pub:latest worker1 Running Running 15 minutes ago
k1dzlwdjogi3 tyk-pro-demo_tyk-gateway.1 tykio/tyk-gateway:latest worker2 Running Running 15 minutes ago
j6j1162d2mlv tyk-pro-demo_tyk-dashboard.1 tykio/tyk-dashboard:latest worker2 Running Running 15 minutes ago
4sm3k3xgcjmf tyk-pro-demo_tyk-mongo.1 mongo:3.2 manager1 Running Running 15 minutes ago
6sjb2gde9cl6 tyk-pro-demo_tyk-redis.1 redis:latest manager1 Running Running 15 minutes ago
etgp6m07ff13 tyk-pro-demo_tyk-pump.2 tykio/tyk-pump-docker-pub:latest manager1 Running Running 15 minutes ago
tddoyri3kcw4 tyk-pro-demo_tyk-gateway.2 tykio/tyk-gateway:latest worker1 Running Running 15 minutes ago
tzpej4dojerm tyk-pro-demo_tyk-dashboard.2 tykio/tyk-dashboard:latest worker1 Running Running 15 minutes ago
Note that for demo purposes in this stack all the services have their ports forwarded. Production must not leave MongoDB and Redis ports open to the world.
It's time to make a demo setup. If running on a manager node created by the docker-machine, first install bash and Python2 on it:
$ tce-load -wi bash
$ tce-load -wi python
Bootstrap the instance:
$ chmod +x setup.sh
$ ./setup.sh
Using 127.0.0.1 as Tyk host address.
If this is wrong, please specify the instance IP address (e.g. ./setup.sh 192.168.1.1)
Creating Organisation
ORG ID: 5a96ca383400ac00016851b6
Adding new user
USER AUTH: 0d2abc2d27be405261867f95136891e6
USER ID: 5a96ca282212f1ff6d7c5df1
Setting password
Setting up the Portal catalogue
Creating the Portal Home page
Fixing Portal URL
DONE
====
Login at http://www.tyk-test.com:3000/
Username: test22014@test.com
Password: test123
Portal: http://www.tyk-portal-test.com:3000/portal/
Alternatively, run the script on another host that can reach the manager node externally.
If the same demo data is used, make sure to add your manager node to /etc/hosts like this:
192.168.99.100 www.tyk-test.com www.tyk-portal-test.com
The manager node acts as a swarm load balancer and discovery service so it can always be used to access all the services.
This concludes the guide and (if nothing's been changed) you should have a docker swarm of 3 nodes, containing 1 MongoDB, 1 Redis container on the manager node, 2 of each dashboard and gateway spread over the worker nodes and 2 pumps across the swarm.
Note that docker-machine VMs only retain the volumes, images, containers and certain docker config files. Everything else will be lost on VM restart unless explicitly saved by the VM provider.