HolyArrs vs the arr stack: should you switch?
HolyArrs folds Sonarr, Radarr, Prowlarr and three more apps into one container. See what 1.0 still lacks, and how to trial 1.0.19 beside your stack safely.
Should you switch from the arr stack to HolyArrs?
HolyArrs is a new app that does the jobs of Sonarr, Radarr, Lidarr, Bazarr, Prowlarr and Seerr in one container, behind one login and one web port. If you have an arr stack that works, try HolyArrs beside it for now and do not switch yet. On 11 November 2026 the project was two days old and was shipping several releases a day. Version 1.0.19 also lacks features that many stacks depend on, such as importing custom formats.
A trial is cheap. HolyArrs only reads your old apps and never changes them. Its import never moves, renames, copies or deletes a media file. This guide covers what HolyArrs replaces, what it still lacks and who should wait. It then shows how to run the pinned 1.0.19 image next to an existing stack without touching that stack.
What does HolyArrs replace?
The HolyArrs README puts it plainly: "A normal media server runs six apps: Sonarr, Radarr, Lidarr, Bazarr, Prowlarr and Seerr." HolyArrs does the work of those six in a single container. It manages your libraries, keeps subtitles in the languages you pick, handles requests and talks to your indexers. Everything sits behind one owner account on port 6262.
It does not replace your download client. The README says: "HolyArrs connects to qBittorrent, SABnzbd and the others. It doesn't replace them." On 11 November 2026 the README listed qBittorrent, Transmission, Deluge, rTorrent, SABnzbd and NZBGet as supported downloaders. rTorrent and ruTorrent support arrived in release 1.0.19. The README describes a built-in downloader as "planned for a later release". On 14 November 2026 the current README advertised 1.1.4; aria2 support arrived in 1.1.0, and the trial below stays pinned to 1.0.19.
It may not replace Prowlarr either, even though Prowlarr is one of the six. HolyArrs calls an indexer feed a source, and its connect docs describe a source as a feed "usually through Prowlarr or Jackett". In 1.0.19, the supported source types are Prowlarr, Jackett, Torznab, Newznab and RSS. Release 1.1.1 added Knaben, TorrentsCSV, SubsPlease and IPTorrents without Prowlarr. Prowlarr pushes its sites to HolyArrs and keeps them current. HolyArrs pulls from Jackett every 6 hours, or when you press Sync. RSS gives the newest items only and cannot be searched. So if your indexers already live in Prowlarr, the simplest path is to keep Prowlarr running and point HolyArrs at it.
Your media server also stays. The README lists Plex, Jellyfin and Emby under media servers, and says your media server "hears about it when it lands". The player is still a separate app, as described in adding Jellyfin to an arr stack.
What HolyArrs still lacks in 1.0
These gaps and disabled features come from the 1.0.19 README and docs as read on 13 November 2026. Each one is a reason for some readers to wait.
- No built-in downloader. You still run qBittorrent, SABnzbd or another client.
- No phone or TV apps. The README says "Phone and TV apps. Not built." It adds that the web app works on a phone.
- Online subtitle search is off by default. Version 1.0.19 includes subtitle sites and online searches, hidden unless the
subtitle_sitesswitch is on. - Music in Explore is unfinished. The README says "Music in Explore is still being built."
- AI connections through MCP are off by default. Version 1.0.19 includes the
ai_appsswitch for Settings > AI apps, the/aipage and its routes. - Custom formats, tags and history do not transfer. The README says "Custom formats, tags and history don't come over." If you use custom formats to control release quality, look in the HolyArrs docs for an equivalent before you plan a move. The import will not carry your rules across.
Since 1.1.1, subtitle sites and subtitle search start enabled. The 1.1.1 changelog also reports old-app formats importing with exact scores, tags selecting each title's Sources, and Profilarr sending formats, profiles and settings. The README and migration docs still say custom formats and tags do not transfer, so check the import report before relying on that newer support. These changes do not apply to the pinned 1.0.19 trial below.
Keeping subtitles and finding subtitles are two separate features. HolyArrs 1.0.19 includes both, but online search is held back by default. To enable it, add HOLYARRS_FLAGS=subtitle_sites to the container's environment and restart it. To enable AI connections too, use HOLYARRS_FLAGS=subtitle_sites,ai_apps. The docs warn that held-back features are enabled at your own risk.
How mature is HolyArrs on 11 November 2026?
The project is very young and changes fast. These figures come from GitHub and Docker Hub, read on 11 November 2026:
- The GitHub repository was created on 9 November 2026 at 23:53 UTC.
- The
1.0.0image was pushed to Docker Hub on 10 November at 00:21 UTC. The1.0.19image followed on 11 November at 10:15 UTC. That is 20 releases in about 34 hours. - The repository showed 31 commits, 88 open issues, 1 open pull request, 762 stars and 35 forks. It was not archived, and the last push was on 11 November at 11:18 UTC.
Twenty releases in a day and a half means bugs get fixed quickly. It also means any upgrade can change how the app behaves. Release 1.0.19 alone fixed four things: Usenet imports, the download counts in the Downloads tabs, the free-space display, and a problem with a shared SABnzbd. A project this new has not yet run through months of upgrades on top of a real library.
Get version information from the GitHub releases page and Docker Hub. When we checked holyarrs.com on 11 November, it still showed a countdown to version 1.0.0 and an old version number in its menu. On 14 November 2026 its menu showed 1.1.4.
What do the size claims mean?
The README says HolyArrs is about 16.1 MB to download, and uses about 140 MiB of RAM at idle with 10,000 titles. It compares the download to the six images it replaces. These are the author's own measurements, using the compressed sizes shown on Docker Hub for linuxserver.io images on amd64, measured on 13 November 2026 (HolyArrs 1.1.3). The RAM measurement is dated 8 November 2026, with 10,000 titles and 55,000 movies and episodes. We did not measure them ourselves.
The data behind this chart
[
{
"label": "HolyArrs (about)",
"download_mb": 16.1
},
{
"label": "The six linuxserver.io images it replaces",
"download_mb": 685.0
}
]The six images download 685.0 MB together. HolyArrs downloads about 16.1 MB. The README calls that "About 43 times smaller." This is a download size, not memory use. On a small VPS the RAM figure matters more. HolyArrs also switches on a low-memory mode by itself when the machine has about 2.5 GiB of memory or less.
What does the licence allow?
The README says HolyArrs is "free and open source under the GNU AGPL-3.0", and GitHub reports the same licence. It also offers a commercial licence, by email, to anyone who wants to build HolyArrs into a closed product. Running your own media library does not involve building a closed product.
Who should switch now, and who should wait?
Wait for a later release if any of these describe your setup:
- You rely on custom formats or tags to pick releases. They do not transfer in 1.0.19; check the newer import support described above before choosing a version.
- Bazarr finds subtitles online for you, and you do not want to enable HolyArrs' held-back subtitle search in 1.0.19.
- Your stack has run for months without problems, and you value that. A project that ships several releases a day cannot promise the same.
- You need a native phone or TV app.
Trying it now makes sense in these cases:
- You are building a new stack, so you have no history or tuned rules to lose.
- You run on a small VPS where six containers use a real share of the memory.
- You want one login and one port in place of six apps and six API keys.
If you do not have a stack yet, the established route is still an arr stack with Docker Compose. The rest of this guide assumes you run that stack.
Run HolyArrs 1.0.19 beside your existing stack
HolyArrs gets its own Compose project and its own config folder. It shares the /data mount that your other containers use. It also joins your stack's Docker network, so it can reach Sonarr and qBittorrent by name. Nothing in the old stack changes.
The steps assume the layout from the arr folder structure and hardlinks guide. That layout is one host tree at /srv/data with torrents/, usenet/ and media/ inside it, mounted as /data in every container. The HolyArrs docs suggest an extra downloads/ level, as in /srv/data/downloads/torrents. Either layout works. Hardlinks need one parent folder, kept on one filesystem and mounted once into each container.
Step 1: create the config folder and find your network
sudo install -d -o 1000 -g 1000 /srv/holyarrs/config
mkdir -p ~/holyarrs
stat -c '%u:%g' /srv/data
docker network ls --filter name=arrstat prints the user and group that own your data tree. That pair should match the PUID and PGID in the compose file below, which is 1000:1000 here. If it prints different numbers, use those numbers instead. How Docker PUID and PGID work explains why they must match. docker network ls should list arr_default if your stack lives in ~/arr. Compose names the default network after the project folder, so a stack in another folder has a different network name. Use the name you see.
The config folder must be on a local disk. The HolyArrs install docs say "Not a network share."
Step 2: write the compose file
Save this as ~/holyarrs/compose.yaml:
name: holyarrs
services:
holyarrs:
image: holyarrs/holyarrs:1.0.19
container_name: holyarrs
environment:
- PUID=1000
- PGID=1000
- TZ=Etc/UTC
volumes:
- /srv/holyarrs/config:/config
- /srv/data:/data
ports:
- 127.0.0.1:6262:6262
networks:
- arr
restart: unless-stopped
networks:
arr:
name: arr_default
external: trueThis file differs from the official example in three places. Each change has a reason.
The tag is pinned. The docs use latest, which moves to every new release. The 1.0.19 tag is one exact version that never moves, so pulling the image cannot upgrade you by surprise. This project shipped 20 releases in two days, so you want to choose when you upgrade.
The port is bound to 127.0.0.1. HolyArrs has no default password. The docs say: "Whoever opens a new HolyArrs first becomes the owner." If you publish 6262 on the public interface, a stranger could reach the setup page before you do. The docs say not to forward 6262 on your router, and to use a VPN or a reverse proxy with HTTPS. On a VPS, binding the port to loopback follows the same advice.
One /data mount replaces four folders. The official example mounts /movies, /tv, /music and /downloads separately. The README warns about the cost: "each mount is its own filesystem inside the container, so every import is a full copy." A single /data mount, the same one Sonarr and qBittorrent use, keeps hardlinks working and gives every container the same paths.
Step 3: start it and check it is healthy
cd ~/holyarrs
docker compose up -d
docker compose ps
docker logs holyarrsdocker compose ps should show the holyarrs container as healthy once HolyArrs responds. If up stops with an error saying that the network arr_default was declared as external but could not be found, the name in compose.yaml does not match your stack's network. Go back to Step 1 and copy the exact name.
Step 4: claim the owner account and save the key
From your own computer, open an SSH tunnel to the loopback port:
ssh -L 6262:127.0.0.1:6262 you@your-vpsOpen http://127.0.0.1:6262 and create the owner account straight away. For permanent access, put HolyArrs behind HTTPS the same way as Sonarr and Radarr behind a reverse proxy.
Next, save the master key. The first start creates /config/holyarrs.key, which HolyArrs uses to encrypt the passwords and API keys it stores. The docs say: "Backups never contain this key. Keep a copy somewhere else." Open Settings, then Backups, and choose Download key kit. Store that file away from the server and apart from your backups.
Step 5: import from your old apps
Open Settings, then Import from other apps. The import can read Sonarr, Radarr, Lidarr, Bazarr, Prowlarr, Seerr and Overseerr, plus qBittorrent and SABnzbd. For each app, enter its address and API key.
Use the container name as the address, for example http://sonarr:8989 and http://radarr:7878, with the ports your own stack uses. Do not use localhost. Inside the HolyArrs container, localhost means HolyArrs itself, so the connection fails. Finding each app's port and API key works the same way as in connecting Prowlarr, Sonarr and Radarr to a download client.
The import has five steps: Read, Choose, Try it, Import, Done. At Try it you see what will be added, and nothing changes yet. Read the "not imported" list at this point. It names each setting that has no match in HolyArrs and gives a reason. For your own setup, that list is the most accurate record of what a move would lose.
After the import, sources, downloaders, alerts and outside lists arrive paused or off. Keys, passwords, tokens and cookies never come over. Type each secret again, check the connection, then turn the item on. If you do not like the result, Undo this import removes what that import added, except items you changed or activated afterwards. The titles leave the library, and the files stay where they are.
Step 6: check that hardlinks work
HolyArrs sees the same /data paths as qBittorrent, so you do not need a path mapping. The docs say "Matching paths need no mapping." In the HolyArrs folder check, a library folder marked No hardlink means downloads and media are on different mounts or filesystems. Every import is then copied. A copy takes longer, and it uses double the space until the downloader deletes its own file.
After a few imports, you can confirm on the host:
find /srv/data/media -type f -links 1 | headA file hardlinked between the downloader and library has a link count of at least 2 while both names remain. Existing apps or other library paths can add more links. This command lists files with exactly one hard link currently present; it does not prove they were copied. A moved file or a previously hardlinked file whose other names were removed can also have a count of 1. For a fresh torrent import whose downloader file still exists, compare the device and inode numbers of both paths with stat -c '%d:%i %h %n' followed by the two file paths. Matching device and inode numbers confirm that they share the same file.
Sharing a downloader between both apps
The docs say HolyArrs "puts its downloads in one category and only watches that one". So it does not pick up the downloads that Sonarr or Radarr started. Both apps can still search for the same title, though. During the trial, decide which app is in charge of each title. One simple rule is to give HolyArrs a few new titles and leave the rest of the library to the old stack.
Upgrading or backing out
To upgrade, read the release notes, change the tag in compose.yaml, and run:
cd ~/holyarrs
docker compose pull
docker compose up -dTo back out completely, stop HolyArrs and delete its config:
cd ~/holyarrs
docker compose down
sudo rm -rf /srv/holyarrsHolyArrs only read your old apps during import, so that step left them unchanged. Media files under /srv/data stay in place.
What to watch before switching for good
If you decide to wait, check these signals on the GitHub repository every few weeks:
- The release pace. Fewer and larger releases usually mean the basics have settled.
- The open issue count, compared with how fast issues get closed.
- Subtitle search and a built-in downloader. Subtitle search starts enabled from 1.1.1; in 1.0.19 it remains behind
subtitle_sites. The built-in downloader is still planned. - The newer custom-format import support, if your release choices depend on it. The 1.1.1 changelog reports it, but the README and migration docs still describe the older limitation.
Tools in this space come and go, and a project can stop as fast as it started. The arr world has seen that before, as covered in what to run now that Huntarr is gone. Keep your old stack runnable until HolyArrs has proved itself. The import leaves that stack unchanged.
FAQ
Does trying HolyArrs change my Sonarr or Radarr setup?
No. During import, HolyArrs only reads the old apps and changes nothing in them. The import never moves, renames, copies or deletes a media file. Imported downloaders and sources arrive paused or off. HolyArrs does not start grabbing anything until you type the secrets again and turn each one on.
Does HolyArrs replace qBittorrent or SABnzbd?
No. HolyArrs does not download by itself. It connects to a download client you already run. On 11 November 2026 the README listed qBittorrent, Transmission, Deluge, rTorrent, SABnzbd and NZBGet. The README says a built-in downloader is planned for a later release.
Why does HolyArrs copy files instead of hardlinking them?
Each separately mounted folder counts as its own filesystem inside the container, and a hardlink cannot cross filesystems. The official compose example mounts /movies, /tv, /music and /downloads separately, so every import becomes a copy. Mount one parent folder, such as /srv/data:/data, in HolyArrs and in your downloader, at the same path and on one disk. The folder check shows No hardlink until the problem is fixed.
What do I lose when I move from Sonarr and Radarr to HolyArrs?
In the pinned 1.0.19 trial, custom formats, tags and old-app history do not transfer. You must type keys, passwords, tokens and cookies again. Settings with no match in HolyArrs appear on a "not imported" list, with a reason for each. HolyArrs 1.0.19 has online subtitle search behind the disabled subtitle_sites switch, and it has no phone or TV apps. For newer releases, check the 1.1.1 changes described above and your import report.
Is HolyArrs free?
Yes. HolyArrs is open source under the GNU AGPL-3.0. The README also offers a separate commercial licence, by email, for anyone who wants to build HolyArrs into a closed product.