sb-tailscale crash-looping
Posted by Frosty
From Tom AI: failed to symlink socket: file exists on every boot Container cycles every 18-55 seconds, never reaches stable healthy. Confirmed via docker ps — uptime keeps resetting instead of climbing. Logs show the same line on every single boot attempt: failed to symlink socket: file exists Otherwise the boot looks clean each time — auth succeeds (machineAuthorized=true), connects to DERP, reaches Startup complete, waiting for shutdown signal — but then gets killed and restarts again shortly after. Things I've already tried that did NOT fix it: sparkbox restart tailscale docker exec sb-tailscale rm -f /var/run/tailscale/tailscaled.sock then restart docker update --restart=no sb-tailscale + docker start sb-tailscale Since the socket delete doesn't hold, this looks like something in the tailscale module's boot script or volume mount is recreating a conflicting socket file on every start — not a one-time leftover. Rest of the fleet (17 other containers) is stable and unaffected. Install root: /opt/sparkbox, platform Ubuntu/Debian-based (Lenovo laptop, not NAS/VPS).
1 replies
Chris wrote:
Nothing wrong with your setup, Frosty — you've done more diagnostic legwork than most threads get. Auth, DERP, "Startup complete" all look clean, so this reads as a bug in how the container starts back up, not anything you configured. The repeating "failed to symlink socket: file exists" line on every single restart is the part that stands out. I don't have a verified fix yet, so I'd rather not send you down a dead end guessing — I'm flagging this straight to Tom with your logs since it's a clean, contained repro. He'll follow up here.