Linux/systemd colibri-bridge packaging + domedog network facts #203
Loading…
Add table
Reference in a new issue
No description provided.
Delete branch "feat/colibri-bridge-linux-packaging"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Records the Colibri TCP-bridge story end to end for the Linux/domedog side, mirroring
packaging/freebsd/colibri_bridge.inon hermes.What's here (
packaging/linux/)colibri-bridge.service— systemd unit runningsocat TCP-LISTEN:9190,bind=$ADDR,freebind → UNIX-CONNECT:$SOCKET,BindsTothe daemon, sandboxed.freebindbinds the tailnet IP beforetailscaledis up (the one improvement over the rc.d version).colibri-bridge.env.example— tunables.COLIBRI_BRIDGE_LISTEN_ADDR=TAILSCALE_IP_REQUIRED(operator fills in viatailscale ip -4at deploy; real 100.x addresses are never committed).colibri-bridge.nft— nftables ruleset for hosts without ufw.README.md— install steps, verified domedog host facts, open questions for the hermes review.Network gate — already live on domedog
ufw allow in on tailscale0 to any port 9190 proto tcpis applied (v4+v6). domedog runs ufw default-deny, so the public side was already blocked and only the tailnet-scoped allow was needed. The nft table is kept for non-ufw hosts only (under ufw itsacceptisn't terminal and itsdropis redundant).Host facts recorded (verified 26.jun.2026)
clawdijavsclawdie; distinct tailnet100.xaddresses (viatailscale ip -4, not committed).cloudpanel.conf, userclp); only 8443 listens. ⚠️ Its admin login is exposed to the public internet — flagged for the host-exposure review.Status
colibri_bridge.inhealth/status bug was fixed+pushed on the hermes side during this work.Updated: scrubbed real Tailscale IPs from all committed files per the no-100.x-in-git policy.
🤖 Generated with Claude Code