One process next to your Docker or Podman socket. It serves the UI on port 9000 and keeps everything in one data directory. Pick the way you run things; each one takes about a minute.
Runs on
Linux amd64 and arm64; macOS and Windows through Docker Desktop or Colima
The installer checks this machine, Docker or Podman, the socket and a free port, then starts FleetOps and waits until it answers. It never installs Docker for you, and asks for sudo only to write outside your home directory. Read it first if you like: install.sh · install.ps1.
Linux and macOS
$ curl -fsSL https://fleetops.run/install.sh | sh
Windows PowerShell, with Docker Desktop
$ irm https://fleetops.run/install.ps1 | iex
Options
Add them after sh -s --, for example curl -fsSL …/install.sh | sh -s -- --port 9100. In PowerShell: & ([scriptblock]::Create((irm https://fleetops.run/install.ps1))) -Port 9100.
Linux, macOS
Windows
What it does
--port 9100
-Port 9100
Use another port. Without it, the first free one from 9000.
--method binary
–
Linux: the verified binary under systemd instead of a container.
--version v1.4.1
-Version v1.4.1
A specific release instead of the latest.
--upgrade
-Upgrade
Move to the newest release, keeping data and port. An Enterprise-image install moves to the one image.
--uninstall [--purge]
-Uninstall [-Purge]
Remove FleetOps; with purge, its data too.
Rather do it by hand? Every method below is the same thing, step by step.
1
Start the server
Docker recommended
The image is about 20 MB compressed, for linux/amd64 and linux/arm64, and runs as a non-root user with no shell. The same command works on Linux, Docker Desktop and Colima.
Try it
$ docker run -d --name fleetops -p 9000:9000 \
-v fleetops-data:/data \
-v /var/run/docker.sock:/var/run/docker.sock \
--group-add "$(docker run --rm -v /var/run/docker.sock:/s busybox:1.37 stat -c %g /s)" \
fleets/fleetops
--group-add grants the group that owns the socket and nothing more. To keep it running, pin the minor version so docker pull brings only patch releases, and restart it with the host:
To run it as a service, save this as /etc/systemd/system/fleetops.service. systemd gives it its own user, the docker group and /var/lib/fleetops for data.
The chart runs one pod with its database on a persistent volume, the UI on port 9000 and HTTPS for agents on 9443. FleetOps then manages the cluster it runs in as its local environment, through its service account (cluster-admin by default; set rbac.clusterRole to edit or view for less).
Without Helm, apply the manifest rendered from the chart's defaults. Ingress, TLS and license options are in the chart's README, and a helmfile starter holds the settings most installs change.
Needs Rust (stable), Node.js 24 or newer and pnpm. make build builds the UI, then the release binary with it embedded. Development guide
Build from source
$ git clone https://github.com/pravinbanjade/fleetops.git
$ cd fleetops
$ make build
$ ./target/release/fleetops --data ./data --bind 127.0.0.1:9000
2
Create the admin account
Open http://localhost:9000 and create the first account. For safety, setup locks if no admin is created within 10 minutes of starting; restart the server to try again. If the page does not load, check the server answers:
Health check
$ curl -fsS http://localhost:9000/api/v1/health
The Docker socket is root on the host. Don't expose FleetOps to an untrusted network without HTTPS (step 4), and create the admin account right away.
3
Add more environments
All four start in Admin → Environments → Add environment. The UI gives you the exact command.
Remote Docker host
Run the docker run command it shows on that host. It starts fleets/fleetops-agentwith a join token that works once and expires in 15 minutes. The agent dials out over HTTPS, so the server needs step 4 first, and the host opens no port.
Docker Swarm
Store the key it shows as a Docker secret on a manager, then deploy the stack file it gives you with docker stack deploy. It runs fleets/fleetops-agent on every node, now and on nodes that join later. Each agent proves its node with the node's Swarm certificate, and an admin confirms the swarm's root CA before any agent is trusted. Swarm guide
Kubernetes, by kubeconfig
Paste a kubeconfig and pick a context. EKS, GKE and AKS sign-in plugins work when the server runs with --kube-exec-plugins=default and the plugin is on its PATH.
Kubernetes, in-cluster agent
Run the kubectl apply command it gives you. The cluster needs no inbound access and the server never holds its credentials. Choose read-only, standard or cluster-admin access. Kubernetes guide
4
Turn on HTTPS
Needed before other machines use FleetOps, and for remote agents. With --tls-bind alone, FleetOps makes a self-signed certificate and keeps it in the data directory:
Behind a reverse proxy, agents still need TLS all the way to FleetOps: pass TLS through, or re-encrypt to the --tls-bind port. For single sign-on, also set FLEETOPS_PUBLIC_URL to the address people use.
5
Enterprise, if you want it
Nothing to reinstall. The same fleets/fleetops runs as Community until you paste a license key in Admin → License, then the Enterprise features turn on. Coming from fleets/fleetops-ee? It is a deprecated alias of the same image: switch the image name on the same data volume, or run the installer with --upgrade. Turn on Enterprise
Keep it running
Upgrade
Data stays in the volume. Releases follow semantic versioning, so minor versions don't break; still, read the release notes and back up before a new major version.
Docker
$ docker pull fleets/fleetops
$ docker rm -f fleetops
# then the same docker run as before
Compose
$ docker compose pull
$ docker compose up -d
Back up
In Admin → Backup, choose Download backup and set a passphrase: one sealed file holds users, environments, roles, settings and the audit log. To move or recover, choose Restore from backup on a new server's first screen. FleetOps can't recover a lost passphrase.
Or copy the data directory while the server is stopped. Keep fleetops.db andsecret.key together; without the key, stored kubeconfigs cannot be read.
Removes the container. The second line also deletes the data for good.
Remove
$ docker rm -f fleetops
$ docker volume rm fleetops-data # deletes all FleetOps data
If something goes wrong
permission denied on /var/run/docker.sock
The container is missing the socket's group. Keep the --group-add line exactly as shown; it reads the right group on Linux, Docker Desktop and Colima. For the binary, run it as a user in the docker group.
port is already allocated
Something, often Portainer, holds port 9000. Publish another one with -p 9100:9000 and open http://localhost:9100.
Setup is locked
Setup locks if no admin is created within 10 minutes of starting. Restart the server (docker restart fleetops) and create the admin right away.
No local environment after setup
FleetOps registers the local environment on first start only if it can reach the socket. Check the socket mount or DOCKER_HOST, then add it in Admin → Environments.
A remote agent stays offline
Agents need HTTPS end to end: start the server with --tls-bind. Behind a reverse proxy, pass TLS through or re-encrypt to FleetOps, allow WebSocket upgrades on /api/v1/agent/connect and keep idle timeouts above 30 s. For a self-signed certificate, mount it into the agent and pass --ca. A join token works once and expires after 15 minutes.
Where are the logs?
docker logs fleetops, or journalctl -u fleetops for the systemd unit. Add -e FLEETOPS_LOG=debug for more detail.
Still stuck? Open an issue with the output of docker logs fleetops.
Configuration
Flags or environment variables; there is no config file. The most used ones are below, and fleetops --help lists them all.
Flag
Environment
Default
What it does
--data
FLEETOPS_DATA
/data
Directory for the database, the secret key and TLS files.
--bind
FLEETOPS_BIND
0.0.0.0:9000
Plain HTTP listen address.
--tls-bind
FLEETOPS_TLS_BIND
off
HTTPS listen address, for example 0.0.0.0:9443.
--tls-cert, --tls-key
FLEETOPS_TLS_CERT, _KEY
self-signed
Your PEM certificate and key. Both or neither.
--docker-host
DOCKER_HOST
/var/run/docker.sock
Docker endpoint for the local environment.
--public-url
FLEETOPS_PUBLIC_URL
unset
Where browsers reach FleetOps. Single sign-on needs it.
--helm-binary
FLEETOPS_HELM
off
The helm used to install and upgrade charts.
--kube-exec-plugins
FLEETOPS_KUBE_EXEC_PLUGINS
off
default allows the EKS, GKE and AKS sign-in plugins.
--license-file
FLEETOPS_LICENSE_FILE
unset
The Enterprise license file; one uploaded in Admin → License wins.