Fixed Docker socket permission bug that could brick your dashboard
Posted by tomspark
SparkBox 1.6.621 ships with a fix for a nasty one I found on a real box that had been stuck for three months: if your Docker socket's group ID changed (common after a NAS OS or Docker update), the dashboard lost all contact with Docker and couldn't recover, even after restart. No error, no visibility — just silent failure. I've also tightened the logging so you'll see exactly when Docker goes dark, and locked down Portainer when you enable public server mode (it was wide open before). The doctor command now audits all your app ports, not just two, so 'locked down' actually means locked down. If your box is already stuck in this state, run sudo sparkbox update on the machine itself or recreate the dashboard container from Portainer — after that it stays fixed. Here's what changed: - Docker socket permission issues no longer silently break the dashboard - Dashboard logs now clearly report when it can't reach Docker - Public server lockdown now closes Portainer (the Docker control panel) - sparkbox doctor audits all app ports, not a subset - One-command recovery for boxes already in a broken state Head to the Updates tab in your dashboard to pull 1.6.621. If you hit any issues, post here in d/sparkbox. --- Update from your SparkBox dashboard's Updates tab, or see the full release notes on the Releases page.