SparkBox patches — Mon Sep 28, 2026
Posted by Chris
SparkBox releases shipped on Mon Sep 28, 2026 (UTC), in order, newest at the bottom. This thread is updated as releases land during the day. Update from the SparkBox dashboard, or run sudo sparkbox upgrade on your SparkBox host. Legend boxes get every release the day it ships; free boxes get it after it has run there for a week. If a problem remains after updating, submit a private support ticket or reply below with the app name, what you expected, and what happened instead. Earlier days: Sun Sep 27, 2026 · Sat Sep 26, 2026 · Fri Sep 25, 2026 · Thu Sep 24, 2026 · Wed Sep 23, 2026 · Tue Sep 22, 2026 · Mon Sep 21, 2026 v1.6.801 — Backups no longer refuse to run because of a media library they leave out - Backups leave your media library out unless you tick it in, and the archive really did leave it out. But the free-space check that runs first measured the whole data folder with the library still in it, so on a PC or Linux install (where the library sits inside SparkBox's own data folder) a backup of your settings was told it needed about the size of your whole library and refused to start. The check now leaves the library out the same way the backup does, so Back up now runs when the settings fit, and still counts the library if you choose to include it. v1.6.802 — The health check no longer warns about Windows' Public network when yours is Private - On a Windows PC, the health check (sudo sparkbox doctor) always showed a yellow warning that Windows might have your network on its Public setting, which blocks other devices, without ever checking. People whose network was already set to Private saw it every time, took it for the problem, and restarted everything to clear it. The health check now asks Windows: it warns only when the network your PC uses is really set to Public, says so plainly when it is Private, and when Windows can't be asked it shows a short note with the one command to check it yourself. v1.6.803 — The health check now tells you when your media drive isn't connected - If your media folder is under /mnt or /media but no drive is actually connected (mounted) there, everything you download quietly lands on the system disk, the new drive never shows up, and the health check (sudo sparkbox doctor) used to show only green ticks. It now says so plainly: that no drive is mounted at your media folder, the one command that shows where each drive is connected, and how to connect the drive and point SparkBox at it. If you keep media on the system disk on purpose, it is a note, not a problem. v1.6.804 — Tom AI no longer turns app names into “[movie title]” labels - SparkBox hides the names of what you download and watch before anything leaves your box. If one line it read happened to call Jellyfin (or another app, or an everyday word) a film title, it then hid that word for the rest of the chat, so a stopped Jellyfin showed up as “[movie title 2]-media” and the logs command Tom AI suggested could not work. App names and everyday words are now hidden only on the line that named them, so container names and commands read correctly again; real film and show names are still hidden everywhere. v1.6.805 — The health check spots a Jellyfin GPU setting with no GPU behind it - If hardware acceleration was switched on in Jellyfin but no graphics chip actually reaches it — always the case on a Windows PC running SparkBox in WSL2, where Jellyfin cannot use the Intel or AMD graphics — every video your TV or phone needs converted fails to play, while others play fine, and the health check said nothing. sudo sparkbox doctor now warns when Jellyfin is set to a GPU it doesn't have and tells you to set Hardware acceleration back to None; on Windows it also says that software transcoding is the right setting there. v1.6.806 — Tom AI explains the risk when you ask it to format a drive - When you asked Tom AI how to format a drive, it refused (rightly — formatting erases everything, and one wrong letter wipes a different disk), but then gave you advice about folder permissions on your media library that had nothing to do with the question. It now says plainly what formatting would erase, gives you one safe check (lsblk -f, which shows every drive, its filesystem and where it is mounted), and explains how to mount a drive and point SparkBox's media folder at it. It still never hands out a format command. v1.6.807 — A failed SSL certificate for your domain now says why - When you set up your own domain for an app and the SSL certificate could not be issued, Nginx Proxy Manager only ever said "Internal Error", and the Custom Domains setup told you to check that Nginx Proxy Manager was running — which it was. The real reason from Let's Encrypt was hidden in a log on the box. The Custom Domains setup now reads that log and tells you what actually went wrong: the domain's DNSSEC is broken, it has no A record pointing at your home, your router does not forward port 80 to the box, something else (like a Cloudflare proxy) answers on port 80, or Let's Encrypt has paused requests after too many tries — each with the one thing to fix. v1.6.808 — Immich's database-login repair now fixes a login that has really gone wrong - If Immich stopped starting because its database was refusing Immich's password, sudo sparkbox repair-immich-db-auth could wrongly report "Immich's on-disk password already matches its database" and change nothing, so Immich stayed down and the repair looked used up. The repair's check was being let in without the password being tested at all. It now tests the password the same way Immich does, so on an affected box it takes a safety copy of the database, resets only the database login to the saved password, and restarts Immich. Your photos and the database contents are not replaced. v1.6.809 — The Cloudflare Tunnel starts again on its own after your box restarts - If your box restarted — after a power cut, a reboot, or a NAS system update — the Cloudflare Tunnel app could stay stopped while every other app came back, so your web addresses stopped working until you pressed Resume on the classic Home page, and the new dashboard did not show that anything was down. The Cloudflare Tunnel now starts again with the rest of your apps after every restart. Update once to get it; your tunnel token and hostname routes are unchanged. v1.6.810 — An app that could not start now says why when you open its log - If an app stopped and would not come back, running sparkbox logs for it could print nothing at all, because the app never got far enough to write a log and the reason sat with Docker instead. Now sparkbox logs shows that the app is not running, the reason Docker gave for not starting it (for example a port another program already uses, or a folder or device that is missing), and says plainly when the log is empty; sparkbox doctor shows the same reason next to an app that never started. Nothing on your box changes; update with sudo sparkbox upgrade to get it.