1006 on root VPS (systemctl --user unavailable)On a fresh VPS install (running as root), OpenClaw appears installed, but channel checks fail with:
Gateway not reachablegateway closed (1006 abnormal closure (no close frame))In many cases, status output also shows:
systemctl --user unavailable$DBUS_SESSION_BUS_ADDRESS and $XDG_RUNTIME_DIR not definedThere are two common causes in this setup:
openclaw gateway restart relies on service manager lifecycle (systemctl --user on Linux user-service mode).gateway already runningport 18789 is already in useUse one mode only. For root VPS environments, tmux mode is the most reliable quick path.
If gateway is flaky/unreachable after install or update, this is the fastest restart pattern:
tmux kill-session -t openclaw 2>/dev/null || true
tmux new -d -s openclaw 'openclaw gateway run --port 18789'
One-line form:
tmux kill-session -t openclaw 2>/dev/null || true; tmux new -d -s openclaw 'openclaw gateway run --port 18789'
If you installed Feishu manually and also have bundled Feishu, remove the extra local copy:
rm -rf ~/.openclaw/extensions/feishu
Set explicit trusted plugin ids:
openclaw config set plugins.allow '["feishu"]'
openclaw gateway stop 2>/dev/null || true
systemctl --user disable --now openclaw-gateway.service 2>/dev/null || true
Install tmux if needed:
command -v tmux >/dev/null || (apt-get update && apt-get install -y tmux)
Start OpenClaw gateway in background tmux session:
tmux kill-session -t openclaw 2>/dev/null || true
tmux new -d -s openclaw 'openclaw gateway run --port 18789'
openclaw gateway status
openclaw channels status --probe
Expected:
RPC probe: okenabled, configured, runningNote for tmux mode:
openclaw gateway status may still show Service: systemd (disabled) and service runtime as stopped/inactive.RPC probe: ok + Listening: 127.0.0.1:18789 + channel probe output.LOG=$(ls -t /tmp/openclaw/openclaw-*.log | head -1)
tail -n 200 "$LOG"
LOG=$(ls -t /tmp/openclaw/openclaw-*.log | head -1)
grep -Ei 'failed before reply|all models failed|no available auth profile|rate_limit|no api key|pairing|allowlist|unauthorized|telegram|feishu' "$LOG" | tail -n 40
tmux attach -t openclawCtrl+b, then dtmux ls1006, gateway unreachable)gateway status probe OK, channel probe OK)root@<vps>:~/.openclaw# openclaw gateway stop 2>/dev/null || true
... Stopped systemd service: openclaw-gateway.service
root@<vps>:~/.openclaw# systemctl --user disable --now openclaw-gateway.service 2>/dev/null || true
root@<vps>:~/.openclaw# command -v tmux >/dev/null || (apt-get update && apt-get install -y tmux)
root@<vps>:~/.openclaw# tmux kill-session -t openclaw 2>/dev/null || true
root@<vps>:~/.openclaw# tmux new -d -s openclaw 'openclaw gateway run --port 18789'
root@<vps>:~/.openclaw# openclaw gateway status
... Service: systemd (disabled)
... RPC probe: ok
... Listening: 127.0.0.1:18789
root@<vps>:~/.openclaw# openclaw channels status --probe
... Gateway reachable.
... Telegram default: enabled, configured, running, mode:polling, bot:@<telegram_bot>, token:config, works
plugins.allow explicit (least privilege plugin loading).camofox-browser and run gateway in tmux when systemctl --user is unavailableplugins.allow is empty and non-bundled plugins may auto-loadinstall.sh + Feishu setup: duplicate plugin id warning spam and gateway health check failureopenclaw update reports ERROR after version changes, then gateway restart fails with systemctl --user unavailable