Shulou(Shulou.com)06/03 Report--

Original articles, welcome to reprint. Reprint please indicate: Reprint from IT people story meeting, thank you!

Docker DockerSwarm cluster environment building (28)

Last time we learned about docker Swarm. This time we started to operate together. We set up a cluster environment for Swarm. We tested service discovery and Load Balancer under three environments together. One was network discovery under overlay under custom, one was service discovery under Ingress, and one was service discovery under Ingress+link. Just like last Mesos, we first set up three service virtual machines. Source code: github.com/limingios/msA-docker swarm branch

server01

docker running state, switch to root user

docker version

Structure of network by default

docker network ls

view IP address

ifconfig

Set up the manager, initialize the manager node, add a work to the swarm, run the following command to add work to the manager.

docker swarm init --advertise-addr 192.168.66.101

server02

Executes commands added to manager.

server03

Executes commands added to manager.

After adding the work of server02 and server03 to server01, view the network

Before, there were only three, one bridge, one host, and one none; there was an extra docker_gwbridge, and one ingress.

Ingress overlay network, this network needs to use docker-gwbridge bridge to connect. if it is

docker network ls

How docker swarm is highly available

Currently there is only one manager, and if the manager node dies, the cluster dies. Docker swarm has a high availability status, that is, server02 and server03, the manager node itself can also provide the service running status, even if our three nodes are managers, we can still run the service. Now try turning all three nodes into manager nodes.

docker node ls

Upgrade server02 to manager. It becomes Reachable, which is the same principle as zookeeper. When a leader hangs, a new leader can be generated by selecting it.

docker node promote docker-swarm-02docker node promote docker-swarm-03

PS: The above completes the highly available docker swarm cluster environment, which is really simpler than others.

Testing Cluster Environment Small Mirror Testing

Create a small mirror and ping www.baidu.com

docker service create --name test1 alpine ping www.baidu.comdocker service ls

```View logs

docker service logs test1

! [](https://upload-images.jianshu.io/upload_images/11223715-f6ac8c36146450df.png? imageMogr2/auto-orientation/strip%7CimageView2/2/w/1240)* nginx environment test `` bashdocker service create --name nginx nginxdocker service ls

Once an old iron asked me, if the container is created and needs to modify the configuration at the time of creation, I told him that I want to make a tag for the current container into a mirror A, and then through the mirror A, generate a new new container and add your configuration in the run, remember that the premise is to delete the original container.

Do you also need to delete service in docker swarm? It doesn't need to be done directly via Docker Service Update.

docker service lsdocker service update --publish-add 8080:80 nginxdocker service ls

There are too few nginx now, only one. If it hangs up, you have to call someone to solve it. There's a way. A few more. You can wait a little longer, time is not urgent, eat and then go.

docker service scale nginx=3docker service ls

The overlay network under Ingress is used by default, and the two previous services cannot be accessed through the service name. You can try it. Custom networks added to docker swarm. Remove nginx and test1docker service rm nginx test1 just created

create a network

docker network create -d overlay idig8-overlay

Create a service specified network

nginx

docker service create --network idig8-overlay --name nginx -p 8080:80 nginx

alpine

docker service create --network idig8-overlay --name alpine alpine ping www.baidu.com

Check the machine where alpine is located, enter the container, ping nginx

This is on server02.

docker service ps alpine

Enter server02, ping container name is nginx, you can ping the discovery

docker psdocker exec -it 387dd735de74 shping nginx

PS: Current network Ingress, access between containers can be accessed by name. Create a dnsadress for each service in a custom network swarm, which must be a custom overlay.

The way of dr.

The description is that the parameters are incorrect, the port cannot be opened under dnsrr, and there is a conflict between them. dnsrr is for access only by name. It is independent unless it joins the overlay network.

docker service create --name nginx-b --endpoint-mode dnsrr -p 8090:80 nginx

If you want the dnsrr network to be accessible, you can add overlay networks for this service

docker service updata --network-add idig8-overlay nginx-bdocker stack

In stand-alone mode, we can use Docker Compose to orchestrate multiple services, while in Docker Swarm through Docker Stack we can complete multi-service orchestration in Docker cluster environment by modifying the existing docker-compose.yml configuration file slightly.

version: "3.4"services: alpine: image: alpine command: - "ping" - "www.baidu.com" networks: - "idig8-overlay" deploy: replicas: 2 restart_policy: condition: on-failure resources: limits: cpus: "0.1" memory: 50M depends_on: - nginx nginx: image: nginx networks: - "idig8-overlay" ports: - "8080:80"networks: idig8-overlay: external: true

Look at the stack command. The top line goes vip Load Balancer instead of dnsrr.

docker stack ls# -c filename group name docker stack deploy -c service.yml test

PS: DockerSwarm's service discovery, Load Balancer.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.