
Nearly a decade ago, in Monitor #Docker #swarm nodes with #WeaveScope, we looked at how Weave Scope could turn a cluster of Swarm nodes into an interactive, real-time topology map of containers, processes, and network connections.
Even after Weaveworks shut down and archived many of its original repositories, Scope’s UI—with live connection graphs, instant container terminals, and CPU/memory sorting—remains hard to beat for quick visual troubleshooting. Recently, I came across a community-maintained resurrection of the project at mario-ezquerro/scope (documented in his write-up, Reviving Open Source Giants: How I Brought Weave Scope Back with Multi-Platform Docker Support), which updates dependencies and adds multi-arch build workflows for x86_64 and arm64.
Instead of a homogeneous 2017 Docker Swarm, however, today’s homelab is a 4-node hybrid fleet running both Docker and Podman across multiple sites connected by Tailscale. Here is how we got Scope running across all four hosts.
Architecture: One Central UI, Four Hybrid Probes

In our setup, one Docker host acts as the central Scope App + Probe (listening on port 4040 over the Tailscale mesh), while the remaining three nodes run lightweight --no-app probes that stream host, process, and container telemetry back to it.
(One practical note from testing: while mario-ezquerro/scope maintains the updated multi-arch workflow in GitHub, its Docker Hub tag list was empty when we pulled, so our x86_64 nodes ran weaveworks/scope:latest cleanly while keeping mario-ezquerro/scope bookmarked for building custom/ARM64 images from source.)
Bridging Podman Nodes via podman.socket
Classic Scope expects /var/run/docker.sock. Two of our four nodes run Podman instead of Docker, which would normally leave container-level metrics invisible.
Enabling Podman’s Docker-compatible API socket solves this cleanly without installing Docker:
sudo systemctl enable --now podman.socket
sudo podman run -d --name=weavescope-probe \
--restart=unless-stopped \
--privileged \
--net=host \
--pid=host \
-v /run/podman/podman.sock:/var/run/docker.sock \
-v /sys/kernel/debug:/sys/kernel/debug \
docker.io/weaveworks/scope:latest --no-app --probe.docker=true <tailscale-hub-ip>:4040
Because podman.socket speaks the Docker Engine REST API, Scope discovers Podman containers and host interfaces seamlessly.
Three Real-World Gotchas We Fixed
- Duplicate
localhostHostnames: Two nodes still hadlocalhost.localdomainas their OS hostname, causing them to collide in Scope’s Hosts view. Giving each host its real static hostname (hostnamectl set-hostname <name>) and passing-h <name>(or--uts=private -h <name>in Podman) immediately separated all four nodes (node-1throughnode-4) in/api/topology/hosts. - Zero Built-In Auth: Scope exposes container stop/restart/shell controls on port
4040with no authentication. Never expose port4040to the public internet—keep it strictly behind Tailscale ACLs or an SSO reverse proxy. - Dashboard Integration: Adding a quick-launch card to our Nginx Homelab Portal turns Scope into a single click whenever we want to inspect container traffic across the mesh.
Key Takeaways
- The 2017 UI still holds up: Comparing our original 2017 post to today, interactive topology graphs remain one of the fastest ways to spot noisy containers and unexpected network links.
podman.socketmakes legacy Docker tools portable: Binding/run/podman/podman.sock:/var/run/docker.socklets Docker-era monitoring probes work across modern Fedora/Podman hosts.- Tailscale replaces complex Swarm overlay networking: Pointing
--no-app <tailscale-hub-ip>:4040across sites gives you a multi-home container map in under five minutes.