Docker Swarm

Docker Swarm lets you cluster multiple Docker hosts into a single, fault-tolerant orchestration platform. Swarm provides built-in load balancing, service scheduling, and high availability by distributing containerized application replicas across manager and worker nodes.

Before you begin: each host must have Docker Engine installed and reachable over the network. One node will act as the swarm manager; the others will be worker nodes. Use a private network or secure connections for production clusters.

A minimalist slide showing the words "Docker Swarm" centered on a white background with a small blue-to-teal underline and "Copyright KodeKloud" in the bottom-left corner.

Initialize the Swarm (on the manager)

On the node you designate as the manager, run docker swarm init. Use --advertise-addr to specify the manager address other nodes should use to join the swarm. The init command prints the docker swarm join command you will run on each worker.

$ docker swarm init --advertise-addr 192.168.1.12
Swarm initialized: current node (0j76dum2r56p1xfn4u1pls2c) is now a manager.

To add a worker to this swarm, run the following command:

    docker swarm join --token SWMTKN-1-35va8b3fi5krpdkefqxgtm... 192.168.1.12:2377

To add a manager to this swarm, run 'docker swarm join-token manager' and follow the instructions.

Copy the docker swarm join command and run it on each worker node to add them to the cluster.

Why use Swarm vs. running containers manually?

You could manually run containers on each host with docker run, but doing this at scale creates operational challenges:

No built-in scheduling or placement logic.
No automatic load balancing across hosts.
No automatic restart or self-healing of failed instances.
Manual, repetitive commands across many hosts.

Example of repeatedly running docker run on separate hosts (illustrative):

$ docker run my-web-server
9f4e2c8a1d7b

$ docker run my-web-server
1a8c0f2e3d4b

$ docker run my-web-server
5e6d7f8091a2

Docker Services — the Swarm abstraction

Swarm introduces Docker services. A service declares a desired state for one or more replicas of an application. The manager schedules service tasks (container instances) onto worker nodes, handles restarts, and balances placement across the cluster. Create a service on the manager and specify replicas:

$ docker service create --name my-web-server --replicas 3 my-web-server
q8z2pgrmpgha9k2qrqzh7yc9

Inspect where replicas are running with docker service ps:

$ docker service ps my-web-server
ID                  NAME                     IMAGE           NODE        DESIRED STATE  CURRENT STATE
abcd1234            my-web-server.1          my-web-server   worker-1    Running        Running 5s
efgh5678            my-web-server.2          my-web-server   worker-2    Running        Running 4s
ijkl9012            my-web-server.3          my-web-server   worker-3    Running        Running 3s

You can pass many of the same options to docker service create that you use with docker run: environment variables, published ports, networks, resource limits, and placement constraints.

$ docker service create --name my-web-server -e MODE=prod -p 80:80 --network web --replicas=3 my-web-server
pwln3y5gx8qhk0tjvc7brmqz4

Common Swarm commands

Purpose	Command
Initialize swarm on manager	`docker swarm init --advertise-addr <MANAGER_IP>`
Get worker join command	`docker swarm join-token worker`
Add a worker (run on worker)	`docker swarm join --token <TOKEN> <MANAGER_IP>:2377`
Create a service	`docker service create --name <NAME> --replicas <N> <IMAGE>`
List services	`docker service ls`
Inspect service tasks	`docker service ps <SERVICE>`
Scale a service	`docker service scale <SERVICE>=<N>`
Remove a service	`docker service rm <SERVICE>`

Best practices and considerations

Run service management commands (create/scale/update) on manager nodes, not on worker nodes.
Use overlay networks for multi-host container networking.
Configure manager quorum and understand leader election for availability.
Use constraints and placement preferences to control where tasks run (e.g., --constraint node.role==worker).
Plan for rolling updates and use docker service update with update parameters to avoid downtime.

Run service management commands (for example docker service create, docker service ps, docker service scale) on a manager node — not on worker nodes.

Further topics to explore

Manager quorum and automatic failover
Overlay networks and encrypted traffic between nodes
Service updates, rollbacks, and healthchecks
Resource reservations, limits, and placement constraints
Integrating Swarm with CI/CD pipelines and logging/monitoring tools

Introduction

Docker Commands

Docker Images

Docker Compose

Docker Engine, Storage

Docker Networking

Docker Registry

Docker on Mac & Windows

Container Orchestration - Docker Swarm & Kubernetes

Conclusion

Docker Run Commands

Initialize the Swarm (on the manager)

Why use Swarm vs. running containers manually?

Docker Services — the Swarm abstraction

Common Swarm commands

Best practices and considerations

Further topics to explore

Links and references

Watch Video

​Initialize the Swarm (on the manager)

​Why use Swarm vs. running containers manually?

​Docker Services — the Swarm abstraction

​Common Swarm commands

​Best practices and considerations

​Further topics to explore

​Links and references

Watch Video

Initialize the Swarm (on the manager)

Why use Swarm vs. running containers manually?

Docker Services — the Swarm abstraction

Common Swarm commands

Best practices and considerations

Further topics to explore

Links and references