Deploy DUMB on Unraid¶
Deploy DUMB from the Community Applications template when it is available, or add it as a Compose stack/container using the maintained settings below. The same persistence and path rules apply to both methods.
Install and check the Unraid container¶
The application template sets up the container. The first import and playback check happens after DUMB and the selected apps have been configured.
flowchart TD
Apps["Apps: find the DUMB template"] --> Review["Review image, persistent paths and user IDs"]
Review --> Apply["Apply the template"]
Apply --> Logs["Follow the container log through startup"]
Logs --> Dashboard["Open DUMB on port 3005"]
Dashboard --> Setup["Complete onboarding and app setup"]
Setup --> Test["Test import and playback"]
Prerequisites¶
- Unraid with Docker enabled
- Community Applications for the template-based path
- A persistent appdata directory and sufficient space for selected services
- A UID/GID that owns the mapped directories
/dev/fuseaccess for workflows that create rclone/FUSE mounts
Install from Community Applications¶
- Open Apps, search for
DUMB, and select the Distributed Unlimited Media Bridge template. - Confirm the image is
iampuid0/dumb:latest(or the release tag you intend to run). - Configure the four persistent mappings shown below.
- Set
TZ,PUID, andPGIDto values that can access those shares. - Keep
/dev/fuse,SYS_ADMIN, and the required security settings when your workflow uses FUSE. - Apply the template and follow the container log through first startup.
The common Unraid nobody:users identity is 99:100, but do not copy those
values blindly. Use the owner of your appdata/media paths, and correct existing
directory permissions before starting DUMB.
Persistent path mappings¶
Use a layout such as:
| Unraid host path | Container path | Purpose |
|---|---|---|
/mnt/user/appdata/DUMB/config |
/config |
Runtime configuration, auth, metrics, jobs |
/mnt/user/appdata/DUMB/log |
/log |
DUMB and managed-service logs |
/mnt/user/appdata/DUMB/data |
/data |
Service-specific persistent application data |
/mnt/user/appdata/DUMB/mnt/debrid |
/mnt/debrid |
Mounts, links, and library trees |
Do not replace the /data mapping with a generic /mnt mapping. DUMB maps
internal paths such as /plex, /altmount, and /postgres_data into the
persistent /data tree.
If you deploy through Compose, start from the current
DUMB Compose file
and replace /home/username/docker/DUMB with /mnt/user/appdata/DUMB.
Open DUMB¶
The maintained deployment publishes the frontend at:
The frontend proxies DUMB API calls under /api. The backend-native 8000
listener remains private by default; do not publish it for ordinary use.
Share library paths with another container¶
Use the same container-visible path in DUMB and the external Plex, Jellyfin, Emby, Sonarr, Radarr, or other consumer:
Then select the curated library produced by your workflow, for example:
/mnt/debrid/riven_symlinks/mnt/debrid/clid_symlinks/mnt/debrid/decypharr_symlinks/mnt/debrid/infinidysk-symlinks/mnt/debrid/combined_symlinks- the AltMount import directory configured for that instance
Do not point a media library at an entire raw provider mount unless that is explicitly the workflow you designed.
Mount propagation¶
Prepare the producer host path as a shared Linux mount before relying on
rshared/rslave. Compose propagation flags alone do not change the host
mount. Check both the producer and consumer mount namespaces.
When DUMB creates a nested FUSE/rclone mount that another container must see:
- DUMB's host bind must be
rshared. - Consumer containers should mount the same host path with
rslave(or an equivalent slave propagation mode). - Restart consumers after changing propagation settings and verify the nested mount from inside each container.
Propagation is unnecessary for ordinary symlink/file trees and when every consumer runs inside DUMB.
Permissions¶
Check the selected IDs and the exact directory before changing ownership:
DUMB's controller prepares service ownership at startup. Do not recursively
chown the entire DUMB tree: private controller state under /config can
intentionally have a different owner from managed applications, and a nested
media mount can expose a much larger tree than expected. Fix only the specific
incorrect host directory after stopping its writers and checking its mounts.
Troubleshooting¶
View the container log from the Docker tab, or inspect from an Unraid terminal:
Do not install ad-hoc tools into the running container; those changes disappear on recreation and can alter the supported runtime. Useful checks are already available through the DUMB UI, logs, and standard Ubuntu utilities in the image.
Common causes of failure are:
- incorrect appdata ownership or mismatched
PUID/PGID - missing
/dev/fuseorSYS_ADMINfor a FUSE workflow - inconsistent paths between DUMB and external consumers
- missing
rshared/rslavepropagation for nested mounts - a host port already in use
Continue with Getting Started, then review Service ports before publishing any additional UI.