Features
Everything here is free on the Community plan, except the rows marked Enterprise, which a license key turns on in the same image.
Live board
The screen reflects Docker's current state without a refresh, and says so plainly when it cannot.
- Dashboard
- What needs attention first, then what is running with live CPU and memory, the events tape, disk usage and networks.
- Event-driven lists
- Container, image, volume and network lists follow Docker events over one websocket. No polling, no refresh button.
- Live stats
- CPU and memory columns with a one-minute sparkline. Only rows on screen subscribe.
- Stats charts
- CPU, memory, network and block I/O for the last two minutes on each container.
- Honest offline state
- A dropped connection or a paused watch is named on the page, with the reason and a retry, never shown as current.
Logs
A first-class log viewer, built to stay responsive past 100,000 lines.
- Virtualized
- Keeps the newest 200,000 lines. Follow mode pauses when you scroll up.
- Search
- Text or regex with highlighting and next and previous.
- Filters
- stdout and stderr toggles, timestamps, wrapping, tail and time-range controls.
- Resume without repeats
- A reconnect or a container restart picks up where the log left off.
- ANSI colour and download
- Colours as the app wrote them. Download the lines shown or the full log.
Containers and terminal
Everything you do with docker on the command line, with the safety rails a shared host needs.
- Container pages
- Overview, Inspect, Env, Mounts, Networks, Stats, Logs, Terminal and Files.
- Terminal
- xterm.js with resize. Picks bash, then sh, or the shell and user you choose. A dropped connection ends the session instead of silently starting a new one.
- Recreate with changes
- If the new container fails to start, FleetOps rolls back to the old one.
- Duplicate
- Copies a container with host ports cleared, so the copy does not clash.
- Create
- A form for the common options, with an advanced JSON override.
- Files
- Browse a container's filesystem read-only and download a single file.
- Bulk actions
- Filter the list, select rows, then start, stop, restart or remove them together.
Stacks, images, volumes, networks
The rest of the Docker surface, with Docker's own names and Docker's own errors.
- Images
- Pull with layer progress, build from a Dockerfile or tar context with a streamed log, tag, history, inspect, remove and prune.
- Volumes
- Create with driver options, inspect, remove and prune.
- Networks
- Create with subnet, gateway, internal and attachable; connect and disconnect containers.
- Compose stacks
- Write a Compose file and its variables in the browser and deploy it to any Docker host, local or through an agent. No docker compose binary needed.
- Builds
- Images built on the target host from the repository or a dockerfile_inline before any container changes. A commit that changes only code redeploys just the services that build it.
- Secrets and configs
- Values stored encrypted and copied into each container before it starts, so nothing is needed on the host. A secret is never shown again once saved.
- Plan and rollback
- Each deploy shows the services it creates, recreates or removes, and a diff of the file. A failed deploy puts the stack back; every deploy is a revision to roll back to.
- Git-backed stacks
- The Compose file from a branch of a git repository, over HTTPS or over SSH with a deploy key, synced by polling or a push webhook from GitHub, GitLab, Gitea or Forgejo. Deploy each commit, or review it first.
- Compose projects
- Projects started outside FleetOps are listed too, with start, stop, restart and down.
Remote hosts, Swarm and Kubernetes
One server, many environments. Each one carries its own colour band and tag.
- Dial-out agent
- fleetops-agent connects out to the server, so a remote host opens no port and never exposes its Docker socket.
- One-time join
- A docker run command with a 15-minute, single-use token. Rotate the key or revoke the host at any time.
- Kubernetes
- Connect by kubeconfig (EKS, GKE and AKS included) or an in-cluster agent. Workloads, services, config, storage, RBAC, events and custom resources.
- Live watches
- Lists and object pages follow the cluster through shared watch streams.
- Safe YAML edits
- Every write is a dry run first, shown as a diff. Server-side apply, with conflicts you can see.
- Helm
- List, install, upgrade, roll back and uninstall releases, each reviewed object by object before it applies.
- Docker Swarm
- One stack file runs an agent on every node, each proving its node with the node's Swarm certificate; an admin confirms the swarm before any agent is trusted.
- Swarm pages
- Nodes, services, tasks, secrets, configs and networks, live, with a visualizer and service logs across tasks. Each node also opens as a Docker host of its own.
- Swarm changes
- Drain, promote and label nodes; create, edit, scale, redeploy, roll back and remove services. A typed confirmation before touching FleetOps' own service, its node or the agents.
- Swarm stacks
- Compose files deployed as Swarm services with the whole deploy block, per-service progress, git sync, rollback and drift marked.
- Suspend and resume
- Scale a workload to zero and bring back exactly as many replicas as before.
Access and audit
Roles and the audit log are free on the Community plan.
- Per-environment roles
- Viewer reads, Operator also changes things and opens terminals, Admin runs the server.
- Namespace and stack scoping
- On Kubernetes, limit a role to some namespaces; on a swarm, to some stacks.
- Audit log
- Every mutating action, every terminal session and every refused request, with who did it. 90 days by default.
- Sign-in
- argon2id passwords, cookie sessions stored as hashes, CSRF on every change, rate-limited logins.
- Audit export
- CSV or NDJSON over any date range, streamed.
- Enterprise
- Teams
- Grant environment roles to a team, and see where each user's access comes from.
- Enterprise
- Custom roles
- Roles built from the permission list, checked on every route.
- Enterprise
- Backup and restore
- One passphrase-sealed file holds the whole server. Restore over a running server or from the first-run screen.
- Single sign-on
- Microsoft Entra ID, Google, GitHub, any OIDC provider, or LDAP. Provider groups decide teams and admins.
- Enterprise
- Scheduled backups
- Daily sealed backups to S3-compatible storage or a local folder, with retention and one-click restore.
- Enterprise
- API tokens
- Scoped tokens for automation, audited like any user.
- Enterprise, planned
Image security
Know which running images carry fixable vulnerabilities, across every host, swarm and cluster.
- Vulnerability scanning
- Trivy scans each image when it arrives and once a day, on Docker hosts, Swarm nodes and Kubernetes clusters. Off until an admin turns it on.
- Enterprise
- Patch queue
- One Security page for the whole fleet, with fixable critical vulnerabilities on running containers first. Each image is scanned once per digest, however many hosts run it.
- Enterprise
- Kubernetes scans
- The cluster pulls the image the way it does for your pods, through node registry access or your pull secrets. Nothing in the image runs.
- Enterprise
- CVE notifications
- Rules send new CVEs to Slack, email (SMTP or Amazon SES), Microsoft Teams, Discord, PagerDuty or a signed webhook. One message per CVE; rebuilds that still carry it stay quiet.
- Enterprise
- Air-gapped mirrors
- Take the Trivy image and its databases from your own registry.
- Enterprise
Interface
Built for an engineer working beside a terminal.
- Command palette
- ⌘K or Ctrl-K jumps to any resource or runs an action on it.
- Keyboard
- j and k walk the attention list, Enter opens logs, / focuses the filter.
- Secret masking
- Values that look like secrets stay masked until clicked.
- Dark panels
- One dark theme tuned for long sessions, with phone and tablet layouts.