SparkBox patches — Fri Oct 16, 2026
Posted by Chris
SparkBox releases shipped on Fri Oct 16, 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: Thu Oct 15, 2026 · Wed Oct 14, 2026 · Tue Oct 13, 2026 · Mon Oct 12, 2026 · Sun Oct 11, 2026 · Sat Oct 10, 2026 · Fri Oct 9, 2026 v1.6.949 — A new SparkBox on a VPS asks for a login on Sonarr, Radarr and Prowlarr - On a box with a public internet address (a VPS), a fresh install turned on "Trust my LAN" for Sonarr, Radarr and Prowlarr, which on a VPS means anyone on the internet could open them with no login. New installs on a public address now ship with logins on; the username and password are on each app's tile on the Home page. Home boxes keep Trust my LAN as before. Already running a VPS? Open the classic dashboard's Settings → General → Sonarr / Radarr / Prowlarr access and switch "Trust my LAN" off. - sudo sparkbox lockdown (and the one-app repair commands) no longer print a long "Found orphan containers … --remove-orphans" warning naming your other apps. Nothing was wrong; do not run that flag — it would remove your apps. v1.6.950 — VPS servers set up before 1.6.949 now ask for a login on the download apps - Servers with a public IP address that were set up before 1.6.949 now ask for a login on Sonarr, Radarr, Prowlarr, Lidarr, Chaptarr and Bazarr after this update, using the admin account SparkBox set up for each one (Settings → Service Passwords, which now has a "Bazarr admin" row). Servers on a home network are unchanged. - If the checkup says a download app opens to the internet without a login (for example after you switched "Trust my LAN" back on), run sudo sparkbox arr-login to switch the login on again. On a server with a public IP address, the classic dashboard now asks before you turn "Trust my LAN" on, and the Security card's "How to lock it down" link opens a Help answer on what SparkBox locks for you and how to close the rest. v1.6.951 — Tom AI spots a library folder the download apps can't save into - When the Media card on your Home page shows a storage problem because Sonarr or Radarr can see your library folder but isn't allowed to save into it, Tom AI now sees the same thing. It tells you so and points you to sudo sparkbox check-media-perms, which checks your library folders from inside the apps and prints the one command that fixes them. Before, Tom AI could report that everything looked fine and hand the problem to support instead. v1.6.952 — Remove SparkBox and keep your movies with one command - New: sudo sparkbox uninstall removes SparkBox (every app, the sparkbox command and the start-at-boot service) and deletes none of your files. It prints where your movies and shows, app settings, passwords and backups are, so you can copy what you want before you delete anything yourself, and how to reinstall into the same folder to pick everything back up. Before, the only full uninstall was sudo sparkbox reset --nuke, which also deletes the media library on a standard Linux or VPS install. reset --nuke and reset --hard now name your media folder before you type YES when it is inside what they delete. - Settings → Your box → overrides: a setting pasted with the lines around it (common when copying from a chat or a support reply) is now saved without them, and a setting written in a way your box could not read is refused with the line number and "Nothing was saved". Before, the page said Saved, the app was not changed, and the box only reported the problem in its log. - The override box's example now uses the app you picked: its real service names (for the Media Center, jellyfin-media) and the memory-limit form SparkBox's own apps use. The old example named jellyfin with memlimit, which does not change the Media Center's Jellyfin. v1.6.953 — A full system disk can no longer lock you out of the dashboard - When the disk SparkBox is installed on was full, signing in to the dashboard failed with "Your password is correct, but this box could not save the login session" and told you to run sudo sparkbox restart, which cannot work on a full disk: no app can be started again until there is room. Your password now signs you in anyway, and the box (and sudo sparkbox doctor) says the install disk is full and to free space there first, instead of suggesting a restart. - Pinchflat no longer fills the system disk. It keeps a small working file for every channel check and every download and never removes them, and those files were stored on the system disk even though your videos go to the media folder. They now go to a hidden folder next to your videos (youtube/.pinchflat-tmp in your media folder), and ones older than a day are cleared each time Pinchflat starts. Your Pinchflat settings, subscriptions and videos are not touched. If your system disk is already full, sudo sparkbox down pinchflat frees what Pinchflat had built up; sudo sparkbox up pinchflat starts it again after updating. v1.6.954 — Frigate security update - Frigate is updated to 0.18.2, a security release from the Frigate team. It fixes two problems that let someone signed in to Frigate (one of them with only a viewer account) run commands on your box, and one where Frigate checked sign-in on some addresses incorrectly. If you use Frigate, update now: your cameras, recordings and settings are kept, and Frigate restarts once on the new version. - This release goes to every SparkBox at once, not only to Legend boxes. v1.6.955 — Tom AI's folder-permission fix ends with a command that works on your box; app updates - When Tom AI's "fix media folder permissions" could not reach the problem because the folder Sonarr or Radarr can't save into already holds your library, it handed you a fixed command that fails on many boxes ("setfacl: command not found" on most NAS boxes, and no effect at all on an NTFS or exFAT drive). It now points you to sudo sparkbox check-media-perms, which tests your library folders from inside the apps and prints the one command that works on your box. Tom AI still never changes who owns the files in your library. - App updates: Nextcloud 35.0.2 (includes security fixes from the Nextcloud team), Homarr 2.6.0, Ghost 6.71, Ollama 0.40.5, n8n 2.44.2, Paperless-ngx 3.3.2 and Stirling-PDF 3.2.0. Your data and settings are kept; each app restarts once on its new version when you update it. v1.6.956 — Ask Tom AI to give an app more memory and it offers the change; a clearer weekly note - When you asked Tom AI to give an app more memory ("raise the memory for Radarr and Sonarr to 2G"), it refused unless the box's checkup had already caught that app running out of memory, so you were left with a refusal instead of the change you asked for. Now asking is enough: Tom AI puts up the change for that app for you to confirm. It still only ever raises an app's memory, never past what your box can spare, and never for SparkBox's own core apps. - The weekly Tom AI note could fill its whole "Needs a look" line with the same problem ("The all update on … failed" three times) and kept listing an app even after a later update of it worked. It now says "Updating all apps failed 3 times this week, last on …" once, and drops an app as soon as an update of it succeeds. The reason for a failed run is under that run in Updates → Update History. v1.6.957 — Installing SparkBox on a server you already use leaves its SSH logins and fail2ban settings alone - Installing SparkBox on a Linux server that had no SSH key set up could switch SSH password logins off, so the next time you tried to log in with your password the server refused it. Now the installer only switches SSH to key-only when you already log in with a key. With no key it leaves SSH as it was and tells you so; add your key, then run sudo sparkbox upgrade to switch to key-only. - Installing SparkBox on a server that already used fail2ban replaced your own fail2ban settings (/etc/fail2ban/jail.local) with SparkBox's SSH rule, so any other protections you had set up there stopped. SparkBox's rule now has its own file, /etc/fail2ban/jail.d/sparkbox-sshd.conf, and your jail.local is left alone; where both cover SSH, your settings win. v1.6.958 — Low on disk space? The checkup now shows what fills Jellyfin's folder and how to free it safely - When the disk SparkBox is installed on runs low, sudo sparkbox doctor now shows how much of it Jellyfin's own folder takes and what is inside: preview images, copies of its database left over from updates that did not finish, Jellyfin's own backups, the library database and artwork, each with its size and the safe step for it. Preview images can be moved next to your videos on your media drive with a Jellyfin setting, which often gives most of the space back. Nothing is moved or deleted for you. v1.6.959 — Nginx Proxy Manager stuck unhealthy over a missing certificate? The checkup now names the one host and the safe fix - When one of your hosts in Nginx Proxy Manager points at a certificate whose files are gone, Nginx Proxy Manager cannot start and its admin page on port 81 never opens. sudo sparkbox doctor (and Tom AI) now says so, names the one host behind it, and gives the steps that keep all your other hosts and certificates: set that one host aside, restart, then give it a new certificate in Nginx Proxy Manager. There is no need to delete Nginx Proxy Manager's folders. Nothing is changed for you.
11 replies
Leo20 wrote:
Is there a way I can switch between the legend update channel (release day updates) vs the basic update channel (weekly)? I want to keep my Production media server a bit more stable? or maybe an Update and Patch channel vs an Update only channel?
Chris wrote:
The current schedule is as described in the post: Legend gets releases on ship day, while the basic tier gets each release after it has run on Legend for a week. I can’t confirm a user-selectable channel switch or an update-only mode from the information available here, so I don’t want to point you to a setting that may not exist. Your request for a choice of update cadence and patch-only updates is clear.
Chris wrote:
Following up with what I could confirm on the box side. There is no channel switch today: the update track follows the licence, so a Legend box always gets the newest release. Your ask for a slower track, or a patches-only track, is noted. What you can do right now to keep production steady: on the production box open Updates, scroll to Auto-Update Schedule, set When updates are found to Just tell me — I'll apply them, and press Save Schedule. From then on nothing installs on its own, neither app updates nor SparkBox upgrades. The Updates page tells you when something is ready, and it only installs when you press Update. Since you already have a test clone, a simple routine works well: let the clone take each release first, use it for a few days, then update production. One thing to know: pressing Update installs the newest release at that moment, so if another one has shipped since, give that one a run on the clone too. The daily "SparkBox patches" posts list every release by date. If a release is marked as a security fix, I'd apply that one to production straight away.
Leo20 wrote:
right now I released the license from my test clone, which also released the license from my production sparkbox. Once that is sorted out, I will re activate my Production and Test Clone, for now they are both on the weekly more stable update path.
Chris wrote:
I'm sorry releasing the test clone's license also deactivated production; please open a private support ticket at https://demox.world/support so Anthony can help with the license issue.
Venom_Cask wrote:
I was curious about this as well, and it's one of the reasons I haven't installed SparkBox yet. Without a definitive answer, I'm leaning toward running it inside its own VM to maintain greater control over my server's security and infrastructure. This isn't necessarily a question of trusting SparkBox or its developer Tom. It's more about taking reasonable precautions, especially with the frequent update cycle and the fact that its long-term reliability is still something I'm evaluating. I want to benefit from everything SparkBox offers without making the rest of my server dependent on it. For now, a dedicated VM seems like the most practical approach to balancing functionality, security, and independence, and that's the route I'm planning to take.
Chris wrote:
That’s a reasonable way to limit the blast radius: a dedicated VM can keep SparkBox’s apps and their changes separate from the rest of your server. It isn’t an absolute guarantee—the isolation still depends on how the VM’s network and storage access are configured—but choosing a VM for that boundary makes sense. For update timing, as noted above, Legend receives releases on ship day and the basic track receives them after a week; this thread doesn’t establish a selectable patch-only channel.
Chris wrote:
A dedicated VM is a sound way to run it, and it's a supported setup: give SparkBox an Ubuntu or Debian VM and install it as normal. It also fits what you describe for another reason: on a regular Linux server the installer changes a few server-wide things (it installs Docker if it isn't there, sets up the firewall rules, installs fail2ban and tightens the SSH login settings), so inside a VM all of that stays on the VM and your main host is left alone. On updates, the definitive answer is in my reply above: there is no channel switch, but you decide when things install. In Updates → Auto-Update Schedule, set When updates are found to Just tell me — I'll apply them and press Save Schedule. After that nothing installs on its own, neither app updates nor SparkBox itself, until you press Update. And a box without a paid licence is already on the slower track: it gets each release after it has run on Legend boxes for about a week, while security fixes go to everyone at once.
Leo20 wrote:
It would be ideal to have a Tester Channel - where updates are frequent and we can submit issue fixes. and a Separate Stable Channel, where it notifies of the commutative updates...or even monthly updates just like HomeAssistant does...restarting the services each time there is an update is extreme. Sometimes right after I update there are already patches and must reboot twice in the same day. Not ideal for long term, but perfectly fine as we help develop this.
Chris wrote:
Thanks for spelling this out. Frequent service restarts are disruptive, and a separate opt-in testing channel alongside a steadier stable channel—with clearer, less frequent update notifications—is useful feedback. I can’t promise a particular schedule or implementation, but the distinction you describe is clear.
Venom_Cask wrote:
I agree with Leo20 on this. I think having separate testing and stable release channels would be a great addition, especially for those of us planning to use SparkBox on production servers. One thing I'd love to see is the ability for Legend subscribers to choose their preferred channel without sacrificing any of their other benefits. Early access is great for those who want to test and provide feedback, but others may prefer a more predictable release schedule. It might also be worth considering consolidated updates for noncritical fixes and features, while allowing important security patches to be prioritized. That could help reduce unnecessary service restarts without slowing down development. I haven't deployed my own instance yet, but this is exactly the sort of flexibility I'd appreciate as I prepare to integrate SparkBox into a larger home server environment. I think the rapid development is encouraging. Having more control over how those changes reach production systems would make it even better.