You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Configurable limit and priority for Docker networks that are reachable from macOS #2772
I'm not sure to label this as a bug or a feature request, but here goes.
Summary
OrbStack appears to make at most 8 active Docker networks reachable from macOS (a host bridge plus a route for each). An active network is one with at least one running container. Further networks are left out, and OrbStack only logs a warning in vmgr.log:
agent:docker | level=warning msg="too many active networks: these won't be reachable from macOS until another network is removed or stops all its containers" networks="[traefik]"
I couldn't find this limit in the docs, there's no setting for it, and the GUI and CLI don't show it. Slots go to networks in the order they become active, so which network loses out depends on the order containers start.
Setup
I run a traefik container on a user-defined network with a static IP (192.168.99.1), which the macOS side relies on. Other projects each have their own Compose network (database, redis, mail, …), so I regularly have more than 8 active networks.
When the proxy's network is the 9th one:
docker network inspect shows the network and the static IP as normal.
Inside the OrbStack VM, the container answers on 192.168.99.1.
From macOS, there's no route to the subnet: route get 192.168.99.1 goes to the LAN default gateway
reverse-proxy.traefik.orb.local resolves to the container's machine-network address (192.168.138.x) instead of 192.168.99.1. That makes the problem harder to spot, because the .orb.local name still works.
Recreating the network only fixes it if a slot happens to be free at that moment. Stopping any other stack fixes it right away, because OrbStack adds the waiting network as soon as a slot frees up.
Requests (any of these would help)
Make the limit configurable, e.g. orb config set docker.max_bridged_networks 16, or explain why it's 8 if it's a hard technical limit.
Let me prioritise networks, for example with a network label such as dev.orbstack.network.always-bridge=true. Networks with the label would always get a slot, and others would be left out first.
Activity