junkerderprovinz

Hi @bthoven, good catch, and thanks for asking! The image rebuilds itself every week, but :latest follows OpenCloud's production line, and upstream's production image is still 7.2.4, while 9.0 is a rolling release. You get 9.0 with junkerderprovinz/opencloud:rolling, the template default, which already runs 9.0.0: edit the container, set the Repository to that tag and hit Apply. That jumps from 7.2 straight to 9.0, so please back up your appdata first and have a quick look at the upgrade notes for 8.x and 9.x: https://docs.opencloud.eu/docs/next/admin/maintenance/upgrade/upgrade-9.x.x

CannonCommanderThe most beautiful file manager you will ever run. Also the only one you need. What is this?Two panels side by side, F5 to copy from one to the other, and an interface so polished you will catch yourself tidying folders just to look at it. CannonCommander runs on your server in the browser, with no desktop squeezed into a container, and on your desk as an app for Windows, macOS and Linux. It knows its way around your server: which files still sit on the cache and which on the array, which disks are asleep, the parity check and the permissions your shares expect. Remote storage comes in through rclone, and a built-in MCP server lets an AI assistant work with your files under your own rights. The folders and files in these pictures are made up. Tabs in both panels, and badges for the files that still sit on the cache Copies run as jobs in the background, and a hard link saves the space where Unraid allows one The preview pane shows the file under the cursor, with the camera data of a photo Highlights & FeaturesTwo panels with tabs, a menu bar, a command search on Ctrl+K and the keys the classic commanders taught your fingers: F3 views, F4 edits, F5 copies, F6 moves, F7 makes a folder and F8 deletes Copies, moves and deletes run as jobs in the background, with speed, time left, pause, a speed limit and a rule for files that already exist, and they pick up again after a restart Deleted files go to a trash on the same share first, and the last actions can be undone Unraid in view: a badge on files that still sit on the cache, a warning before a copy wakes a sleeping disk, the parity check in the sidebar, and Set permissions the Unraid way to hand files to nobody:users Nine views per tab, quick look on the space bar and a preview in the other panel Folder compare and sync, file compare, duplicates, disk usage as a tree or a treemap, batch rename with patterns, archives, checksums, split and join A viewer and editor for pictures, video, audio, PDF, Markdown, code, Office files, e-books and 3D models, and an Office server of your own can edit Office files SFTP, SMB, WebDAV, S3, the big cloud drives and everything else rclone speaks, mounted inside the container so the server and other containers never see them Share links with a password, an expiry date and download or upload limits, and a WebDAV server to mount your files as a drive Users, groups and access rules, two-factor sign-in, passkeys, API tokens and single sign-on through OIDC An MCP server for the AI assistants you already use: each key gets its owner's rights, reads only until you allow writing, and never deletes anything for good A desktop app for Windows, macOS and Linux with the same interface, on Linux also as a Flatpak, from the releases page 42 languages, light and dark, and your own accent colour Images: junkerderprovinz/cannoncommander:latest on Docker Hub and ghcr.io/junkerderprovinz/cannoncommander:latest, for amd64 and arm64 RequirementsUnraid 6.10+ On the first start the container writes a password for the user admin into its log (Docker tab, CannonCommander icon, Logs). Sign in with it and pick a new one of at least 12 characters Keep the WebUI on your own network. To reach it from outside, put it behind a reverse proxy with https or use a VPN such as WireGuard or Tailscale Remote storage needs /dev/fuse and the SYS_ADMIN capability, and the template sets both for you Spinning disks down and Unraid notifications are off by default. To switch them on, set CC_UNRAID_SOCKET to true and map the paths the template describes Bugs, Problems, WishesSomething not working, a question or an idea? The best place for it is an issue on GitHub: github.com/junkerderprovinz/cannoncommander/issues. Everything stays in one place there, and nothing gets lost in a long thread. For a bug, whether you post it there or here, these help: Unraid version (Settings → System Information) Image tag (latest or a pinned version), or for the desktop app your system and the app's version Output of docker logs --tail 200 CannonCommander What you did, what you expected and what happened instead For remote storage, which kind it is (SFTP, SMB, S3 and so on) CreditsCannonCommander started from the code of FileBrowser Quantum by Graham Steffaniak and contributors, which itself grew out of File Browser. Both are under the Apache License 2.0, and their notices are kept in the repo's NOTICE file. Thank you to everyone who worked on them before us. Remote storage runs on rclone, and the image brings ffmpeg, MuPDF and 7-Zip for video, PDF and archives. If CannonCommander has earned a place on your server, toss a coin to your knight: you can buy me a coffee. Self-hosted, AGPL-3.0 licensed: you run it, you own your data and the responsibility.

Hey @castores Thanks, and not silly at all. The version you had only tried BombVault's own SSH key, never your password ("attempted methods [none publickey]"). 10.3.0 and later fix that. After updating, open the storage location, type the sub-account's password under Account password and press Add key. BombVault uses the password once to put its key on the Storage Box and doesn't keep it. Your server and user name are right. Just make sure SSH support is switched on for the sub-account in the Hetzner Console. If it still fails, post the new error here.

Hi @blaess, thanks for the report! Changing the region itself works. What you describe usually means garage.toml is no longer valid TOML after the edit, often because the quotes went missing: it has to be s3_region = "us-east-1". Garage can't start then, and the log shows "TOML decode error". With version 1.0.2 you don't need to edit the file. Update the container and enter us-east-1 in the new S3 Region field (if your template doesn't show it, add a variable S3_REGION). That also fixes the broken line. Your client has to use us-east-1 too. Cheers

TowerTorchYour server's UI on fire. Your disks stay cool. What is this?TowerTorch gives your whole Unraid server a fully revamped, modern interface, without replacing a single page. Dashboard, Main, Shares, Docker, Plugins, VMs and Settings get one clean, consistent look: tiles or lists, cards with switches instead of checkboxes, small animated pictures for the disks, the array and the mover, one accent colour or a rainbow, light and dark. Everything stays where Unraid put it, so nothing you know moves. Underneath, it looks after your containers: a start plan starts each one only once the container it depends on is actually ready, not just running, with per-container limits, a watchdog and schedules on top. A small Go daemon on a local unix socket does the work, and nothing leaves your server. Highlights & FeaturesDocker, Main, Shares, Plugins and VMs as lists or tiles; disk tiles only mention SMART, errors and rates when there is something to say Small pictures: the array's fill level as a vessel, platters that turn while a disk works, data flowing through SSDs, the mover and a parity check drawn while they run, all following your motion setting One accent colour or a rainbow, light and dark, shapes and motion levels, and a master switch that turns all theming off and keeps only the orchestration A start plan with dependency-ordered stages, so an app waits until its database is really ready; the plan window draws each chain lighting up in start order Readiness probes (running with grace, TCP port, HTTP, a log line or a command inside the container), each with its own failure policy A watchdog that restarts a crashed or unhealthy container, capped per hour, and schedules to start, stop or restart at fixed times per weekday Idle stop, inspired by ContainerNursery: a container that stays idle on CPU and network gets stopped; busy and host-network containers are skipped Live CPU, RAM and pinning limits through docker update without a restart, with P/E-core detection on Intel hybrid CPUs Upload and download caps per container, applied inside its own network namespace and read back live in the editor RequirementsUnraid 6.10+ Docker, and the VM manager if you want the VMs page Install from Apps, or under Plugins, Install Plugin with https://raw.githubusercontent.com/junkerderprovinz/towertorch/main/plugin/towertorch.plg Coming from CannonadeCommand? Install TowerTorch the same way. It takes over your settings, start plans and limits and removes the old plugin. Bugs, Problems, WishesSomething not working, a question or an idea? The best place for it is an issue on GitHub: github.com/junkerderprovinz/towertorch/issues. Everything stays in one place there, and nothing gets lost in a long thread. For a bug, open Settings → Utilities → TowerTorch → Info and press Bug report. TowerTorch saves a file with the versions, your settings and start plan and the end of the engine log, with passwords and other secrets taken out. Attach it, whether you post there or here, together with: The page and the action you were on (for example Docker → bandwidth gear → save) Browser and operating system A screenshot, if it is about how something looks CreditsBuilt in Go for Unraid. The idle stop is inspired by ContainerNursery. If TowerTorch lights up your Unraid, you can buy me a coffee. Self-hosted, AGPL-3.0 licensed: you run it, you own your server and the responsibility.

I don't know of a public sync status page. The closest thing is the app's page on ca.unraid.net (https://ca.unraid.net/apps/krusader-1gjxcb9125o7ip), which shows a Last Updated date and the template fields, including the language list. Right now it still shows the old template with Last Updated 2026-09-29, so CA hasn't picked up the new one yet. The change itself is already in the template file on GitHub: https://raw.githubusercontent.com/junkerderprovinz/unraid-apps/main/krusader/krusader.xml CA syncs on its own schedule and I can't speed it up. Until the new entries show up there, you can keep en_gb in the language field as you did.

@Revan335 You don't need separate mounts. The template maps the host path /mnt to /storage with the access mode Read/Write - Slave (rw,slave), so all of those show up as subfolders: /storage/user, /storage/cache, /storage/disks, /storage/remotes, /storage/disk1 and so on. Read/Write is needed for copying, moving, renaming or deleting. With read-only, Krusader can only browse. Slave matters for /mnt/disks and /mnt/remotes: Unassigned Devices mounts those after the container has started, and without slave the container keeps seeing an empty folder until it restarts. One warning that applies to any file manager on Unraid: don't copy or move between a user share path and a disk or cache path of the same share, for example from /storage/disk1/Movies to /storage/user/Movies. Both point at the same files, and that can truncate or delete them. Stay inside /storage/user, or inside the disk paths.

Hey @Revan335 Krusader handles this with a panel profile: 1. Open /storage in both panels, plus any tabs you want to keep. 2. Window → Profiles (Alt+Shift+L) → Add new entry, and give it a name, for example storage. 3. Settings → Configure Krusader → Startup, set Startup profile to the new entry, then OK. From then on Krusader opens with that profile on every start. The default, <Last session>, brings back the folders that were open when Krusader last closed, so if that's all you need, you can skip the profile.

Thanks for the kind words, and for tracking this down @FrequencyLost! Good catch: en maps to US English, and the US date format comes with it. The container already knew en_gb, the dropdown just didn't offer it. The template now has en_gb = English (UK), plus pt_br (Brazilian Portuguese) and zh_tw (Traditional Chinese), and the description says that the language also sets the date format. Community Applications should pick up the new template within a few hours. Your edited template keeps working as it is. The container only reads the code at the front, so your en_gb = UK English does the same thing as the new entry.

With KluthR's okay, a short note for anyone here who is looking at alternatives: I maintain BombVault, a backup and restore app in CA. It backs up containers together with their templates, VMs, the flash drive and ZFS datasets with restic, so the backups are deduplicated and encrypted, and it can copy them off-site. If you want to switch, BombVault can import your existing Appdata Backup archives as restore points, so your old backups come along. A container needs one BombVault backup first, so the restore knows its definition. Good to see chodeus working on the next update here, so Appdata Backup keeps going too. Pick whatever suits your setup. BombVault support thread: https://forums.unraid.net/topic/199509-support-junkerderprovinz-bombvault/

I maintain BombVault, a backup and restore app for Unraid in CA (containers, VMs, flash), so the built-in backup in Unraid 8 matters a lot to me. If someone from the team can share a few details: 1. What does it back up? Appdata, Docker templates and compose stacks, VMs, the flash drive, whole shares? 2. Is data encrypted before it leaves the server, and what engine runs underneath? 3. Will there be an API or hooks, so other tools can start a backup or read its status? Knowing this early would help me make BombVault fill the gaps rather than duplicate what ships with Unraid.

Hi @Mr.win7, thanks for trying it and for the detailed report! You found a real gap. StrawDroid rendered on the CPU by default, and passing /dev/dri alone didn't change that. v1.3.1 fixes it: EMULATOR_GPU now defaults to auto, so the emulator draws on whatever card you pass in. For your NVIDIA card: 1. Set EMULATOR_GPU to auto. An existing install keeps the old swiftshader_indirect value from its template. 2. Add --runtime=nvidia to Extra Parameters. 3. Set NVIDIA_VISIBLE_DEVICES to all. The updated template has a field for it; if yours doesn't, add it as a variable. This needs the Nvidia Driver plugin. Once it has started, the READY block at the end of the log has a Drawn: line, and it should name your card. If it says SwiftShader or llvmpipe instead, it's still on the CPU, and I'd like to see that log. About Redroid: it's faster because it isn't an emulator at all; Android runs straight on the host kernel. StrawDroid is built as a test rig that behaves like a phone (battery, sensors, Doze), so if you mostly want to use apps, Redroid may still suit you better. Keep in mind that Redroid only renders in software on NVIDIA cards. Thanks for offering to test, I'd love to hear how it runs on your card. For bug reports or feature requests, GitHub is the best spot, I see those fastest there: https://github.com/junkerderprovinz/strawdroid/issues

Hi @Moll, yes, that's what it means. Set up the way the guide describes, Config and Data sit on the SSD pool (/mnt/cache/appdata/opencloud/...), and uploads go straight to the array disk you set as Files. Nothing stops on the cache first and the mover isn't involved. Writing to the array is slower than the SSD, mostly because of parity, but with a few files you'll hardly notice. Since you only keep a few files, there's an even simpler option: leave Files empty. Then everything stays in Data on the SSD, and 256 GB goes a long way. Parity doesn't cover it there, so back up the appdata folder, for example with the Appdata Backup plugin. About /mnt/user: OpenCloud has no database and is constantly writing small files. During a big sync, Unraid's share layer can stop responding, and when it does, every container on /mnt/user hangs too until you reboot. With light use you might never run into it, but when it happens it takes the whole server down with it, so I can't recommend it. If a path still goes through the share layer, the container log warns you at every start. The bigger question for you is calendars, contacts and to-dos. OpenCloud doesn't do CalDAV/CardDAV itself. It passes those requests to a separate Radicale server and sits in front of it as a proxy (https://docs.opencloud.eu/docs/admin/configuration/radicale-integration). The template doesn't set that up, so out of the box you only get the file side. If calendars are the main reason you'd switch, keep Nextcloud for them for now or run Radicale on its own.

KnightLoaderDownloads everything, kneels to nothing. What is this?KnightLoader is a download manager for hosters, debrid services, torrents, Usenet, videos and streams. Hoster links go through the debrid service you already pay for, torrents to the built-in client or to a debrid service, .nzb files to your own Usenet servers, and videos and streams to yt-dlp. Whatever is left goes to a headless JDownloader that KnightLoader sets up by itself inside the container. The image brings yt-dlp, ffmpeg and the Java runtime along, so there is nothing to install beside it. The same program also runs as a desktop app for Windows, macOS and Linux, and an Android app and a browser extension send it links from wherever you are. Highlights & FeaturesLinks wait in the collector first. It checks each one and shows what will fetch it, and Packagizer rules sort them into packages and folders as they arrive Instances, the Android app and the browser extension find each other with twelve words, across networks, without an account, a domain or a port forward. The relay can be the project's, your own ParleyPort, or an instance that is reachable from outside Torrents through the built-in client or a debrid service, and Usenet from your own servers with par2 repair or through TorBox and Premiumize.me Sonarr and Radarr see it as qBittorrent or SABnzbd Links come in from Click'n'Load, the browser extension, the phone's share sheet, a watched folder, and DLC, CCF and RSDF container files When a download finishes, archives are unpacked without outside tools, checksums are checked, files land in folders built from templates, and a media library can be told to rescan Coming from JDownloader? KnightLoader takes over your accounts, rules, archive passwords and the open download list, with a preview first A password, two-factor authentication and passkeys under Settings, Security, and an app key that seals the secrets in the appdata folder 42 languages Images: junkerderprovinz/knightloader:latest on Docker Hub and ghcr.io/junkerderprovinz/knightloader:latest, for amd64 and arm64 RequirementsUnraid 6.10+ Keep the WebUI on your LAN until you set a password under Settings, Security: until then anyone who can reach the port can use it An app key is optional but worth it. Run openssl rand -hex 32 in a terminal and paste the result into the App key field (KL_APP_KEY). Keep it with your other passwords, because KnightLoader does not start on sealed data without the same key The template runs the container as nobody:users (--user 99:100) with UMASK 000, so the account you use over SMB can move, rename and delete what KnightLoader downloads Bugs, Problems, WishesSomething not working, a question or an idea? The best place for it is an issue on GitHub: github.com/junkerderprovinz/knightloader/issues. Everything stays in one place there, and nothing gets lost in a long thread. For a bug, whether you post it there or here, these help: Unraid version (Settings → System Information) Image tag (latest or a pinned version) The file from the Bug report button under Settings, Info: the build, the settings with every password removed and the latest log lines If the container does not start, the last lines of its log (Docker tab, KnightLoader icon, Logs) What kind of link it was (hoster, debrid, torrent, Usenet, video) and what you expected against what happened CreditsDownloads run on the engine of Gopeed, videos and streams on yt-dlp and ffmpeg, and JDownloader takes the hoster links nothing else claims. The RSDF and CCF readers follow JDownloader's GPL-3.0 container plugins. If KnightLoader has earned a place on your server, toss a coin to your knight: you can buy me a coffee. Self-hosted, AGPL-3.0 licensed: you run it, you own your data and the responsibility.

Hi @MisterSparkle, thank you for the detailed report, the log showed exactly what happens. This is a bug in BombVault and not in your setup: a timer that stops a backup after two hours without progress kept running during the off-site copy, which reports no progress to it. Every copy that needed more than two hours after a backup was cancelled, and the rclone errors in your log are only that cancellation. Version 9.9.3 fixes it. 1. Unraid web UI, Docker tab: click Check for Updates at the bottom, then apply update in the BombVault row. 2. In BombVault open Settings, Off-site and click Replicate now in the Containers section. The copy runs in the background. 3. The first copy to your SFTP server can take many hours. The Dashboard shows that it is running, and "Off-site replication failed" should not come back. Replicate now was never affected, so step 2 already works on your current version if you want to start right away. If nothing arrives on the SFTP server after the update, please open an issue on GitHub and paste the new log: Docker tab, click the BombVault icon, Logs, select everything with Ctrl+A, copy with Ctrl+C, and paste it into the issue between two lines of three backticks. https://github.com/junkerderprovinz/bombvault/issues