Summary

After running openclaw update from the CLI, the update appeared to install the new package but failed while refreshing the gateway service environment from the updated install. A full server reboot recovered the gateway.

Environment

• OpenClaw after update: 2026.5.26 (10ad3aa)
• Platform: Linux
• Node: 22.22.2
• Install/update channel: stable / pnpm-managed install
• Gateway service mode after recovery: systemd user service

Error

The CLI update path printed an error like:

Restarting service...
Failed to refresh gateway service environment from updated install:
Error: updated install refresh failed (/usr/lib/node_modules/openclaw/dist/index.js):
Gateway install failed:
Error: systemctl enable failed:
Failed to enable unit: Unit file openclaw-gateway.service does not exist.

After rebooting the server, the gateway came back normally.

Expected behavior

openclaw update should refresh/restart the correct gateway service after updating, or at least detect whether the installation is using a user systemd unit versus a system-level unit and call the matching systemctl --user / system-level command.

Actual behavior

The refresh/install step attempted to enable openclaw-gateway.service but failed because the unit file was not found in the scope it used. The gateway was left in a state where manual server reboot was needed to recover.

Why this might be happening

This looks like a service-scope mismatch during the post-update refresh path:

• The healthy recovered state uses a systemd user service.
• The update refresh path appears to call a systemctl enable path that cannot find the expected unit.
• The error message does not tell the user whether it checked user units, system units, or a stale/legacy service name.

Suggested fix

• During update refresh, detect the active gateway service scope and unit path before enabling/restarting.
• Prefer systemctl --user when the active gateway is a user service.
• If the unit is missing, run the same install/repair path used by openclaw doctor or provide a precise command for recovery.
• Make the error actionable, e.g. include which systemctl scope was used and which unit paths were checked.