- Automated Media Server 👻
- About
⚠️ A Note About Torrent Clients- Network
- Installation
- Configuration
- Thank You
- Future Plans
_ _
__ _| |__ ___ ___| |_
/ _` | '_ \ / _ \/ __| __/
| (_| | | | | (_) \__ \ |_
\__, |_| |_|\___/|___/\__|
|___/ / _ \
(¯\| o (@) |/¯)
\_ .___. _/
/ !_! \
/_.--._.--._\
This is an automated media server set up in docker containers via docker-compose. The goal of this project is to automate as much of the installation and configuration as possible while maintaining simplicity of the setup (no ansible playbooks or fancy bash scripts) to avoid making it a complete black box.
The end result of this setup is a media server with the following components
plex
filebot
- with a web interface for triggering
amc
script
- with a web interface for triggering
transmission
- configured to call
filebot
container on torrent completion - with
combustion
web UI
- configured to call
nzbget
- configured to call
filebot
container on nzb completion
- configured to call
sonarr
radarr
jackett
bazarr
tautulli
- configured to look at plex server logs
portainer
watchtower
nginx + letsencrypt
- for reverse proxying to reach your services at *.domain.tld
heimdall
The high-level steps for setup are as follows
- Install docker and docker-compose
docker-compose up
- add an indexer to
jackett
- configure
sonarr
/radarr
to use thejackett
indexer - configure
sonarr
/radarr
to use annzb
indexer - configure
sonarr
/radarr
to usetransmission
as their download client - configure
sonarr
/radarr
to usenzbget
as their nzb client - configure
nzbget
with your newsgroup provider - add libraries to
plex
Once these steps have been completed, it is possible to add a TV Show or Movie to sonarr
/ radarr
, have it automatically download them when available, and then it will be automatically copied into your plex
libraries. Post-installation, this should be a fully-automated media server. The transmission
container will automatically clean any files older than 30 days (configurable).
📓 This has only been tested on Ubuntu 18.04 LTS but it should work just fine on other linux distros. MacOS and Windows are unsupported. If you test it on MacOS or Windows and it works, let me know!
I've removed the cofiguration for qbittorrent
. If you want to keep using it, feel free to search through git history to recover the configuration, but it's no longer supported and has not been updated to use the new filebot
container.
Each service is available on its own ports:
Service | Port |
---|---|
transmission | 5656 |
nzbget | 6789 |
filebot | 7676 |
sonarr | 8989 |
radarr | 7878 |
bazarr | 6767 |
jackett | 9117 |
plex | 32400 |
portainer | 9000 |
heimdall | 8888 |
netdata | 19999 |
📓To reach plex
, append /web
to the address e.g. 192.168.1.11:32400/web
All of the services are running in the default docker-compose network. From within services, they can access each other via their <service_name>:<port>
as defined in docker-compose.yml
. These are NOT the ports you've configured in your .env
file. Those ports are mapped to the host device's network, but the services are operating within the network that docker-compose sets up, so they will not respect the rules forwarding ports to the host network.
Service | Port |
---|---|
transmission | 5656 |
nzbget | 6789 |
filebot | 7676 |
sonarr | 8989 |
radarr | 7878 |
bazarr | 6767 |
jackett | 9117 |
plex | 32400 |
tautulli | 8181 |
heimdall | 80 |
netdata | 19999 |
git clone https://github.com/ghostserverd/mediaserver-docker.git
cd mediaserver-docker
cp .env_sample .env
id $USER # save the result of this for building your .env file below
Modify the .env
file to specify the directory configurations. Note that these directories are for the host machine. They are mapped to various locations inside of their container by the docker-compose
file.
The .env_sample
file has some notes about the various configuration options. There are additional descriptions of each service's configuration below.
A good directory setup looks something like this
CONFIG_DIR=/some/directory/media_server/config
DOWNLOAD_DIR=/some/directory/media_server/downloads
MEDIA_DIR=/some/directory/media_server/media
TV_DIR=/some/directory/media_server/media/TV Shows
MOVIES_DIR=/some/directory/media_server/media/Movies
BASE_DIR=/some/directory/media_server
Note that there is a base volume mount that holds both media
and downloads
/some/directory/media_server
Configuring a single mount that contains both downloads
and media
allows filebot
to take advantage of hardlinks
when renaming and moving files. Hardlinks are much quicker than actually copying the file, but they require that the source and destination be on the same volume in order to work.
variable | description |
---|---|
CONFIG_DIR | where configuration for each of the services will live. You'll end up with multiple directories in here, one for each service. If you move this directory to a different volume with a different instance of the whole media server, it should retain your various configurations. |
DOWNLOAD_DIR | where various services will download files to. |
MEDIA_DIR | where your media will be copied to. |
TV_DIR | where your TV shows will be placed by filebot on download completion. It should be a subdirectory of MEDIA_DIR |
MOVIES_DIR | where your TV shows will be placed by filebot on download completion. It should be a subdirectory of MEDIA_DIR |
BASE_DIR | a shared directory that houses media and downloads directories to be used for hardlinks |
variable | description |
---|---|
PUID | the unix UID that will be passed to the various services. It can be discovered by running id $USER on the host machine. |
PGID | is the unix GID that will be passed to the various services. It can be discovered by running id $USER on the host machine. |
Fill out the rest of the configuration options before deploying. See each individual service's section for configuration details.
docker-compose up
Append -d
to run in detached mode. The first time you run a service, it is probably a good idea to not run in dettached mode (i.e. DON'T append -d
) so you can watch all of the logs for issues.
variable | description |
---|---|
JACKETT_PORT | the port that jacket will listen on |
<server-ip>:9117
jackett
should be configurable the same as any other installation of it. Feel free to skip these steps if you know how to configure jackett
already.
- Add an Indexer
- Click
+ Add Indexer
- Search for your desired tracker
- Click on the 🔧icon next to your desired tracker
- Sign in with your account information
- Click
Okay
- Click
- You should see
Successfully configured <tracker>
and a new entry for your tracker - Copy the
API Key
from the top right corner and save it somewhere - Click the
Copy Torznab Feed
on the tracker you just added and paste it somewhere to save it
variable | description |
---|---|
SONARR_PORT | the port that sonarr will listen on |
<server-ip>:8989
sonarr
should be configurable the same as any other installation of it. Feel free to skip these steps if you know how to configure sonarr
already.
transmission
instead of the IP address when configuring the download client, as well as jackett
instead of the IP when setting up your indexer. This is because this uses docker-compose networking which means each service is accessible at the name of the service, rather than the host or even local IP address.
Step-by-step for those who need it
-
Add an Indexer
- Click on the
Settings
button at the top - Click on the
Indexers
tab - Click the big
+
symbol - Click the
Custom
button in theTorznab
section - Configure your
Torznab
feedName
: the name of this indexer (doesn't matter, just name it the name of your tracker)URL
: theCopy Torznab Feed
url fromjackett
that you saved earlier⚠️ It is necessary to replace the IP withjackett
http://jackett:9117/api/v2.0/indexers/<indexer>/results/torznab/
instead ofhttp://192.168.1.11:9117/api/v2.0/indexers/<indexer>/results/torznab/
API Key
: theAPI Key
fromjackett
that you saved earlier
- Click
Test
to verify that it is configured properly - Click
Save
- Click on the
-
Add a Download Client
- Click on the
Settings
button at the top - Click on the
Download Client
tab - Under
Completed Download Handling
toggleEnable
from Yes to No- We'll be using
filebot
to handle our completed downloads
- We'll be using
- Click the
Save
button at the top right - Click the large
+
button - Click on
transmission
- Configure your Download Client
Name
: whatever you want; probablytransmission
Host
:transmission
Port
:5656
or whatever you have set forTRANS_WEBUI_PORT
in your.env
fileUsername
:admin
or whatever you have set forTRANS_WEBUI_USER
in your.env
filePassword
:adminadmin
or whatever you have set forTRANS_WEBUI_PASS
in your.env
fileCategory
:sonarr
- this allows transmission GC to properly remove torrents after seed limits exceeds
- Click
Test
to verify that it is configured properly - Click
Save
- Click on the
-
Add Some TV Shows
- Click on the
Series
button at the top - Click
+ Add Series
- Start typing in the search bar. It will search automatically when you stop typing
- Configure the download path (you should only have to do this on the first show you add)
- Click the dropdown that says
Select Path
- Click
Add a different path
- Click the 📁button on the right of the modal
- Click
tv
# final path should be/tv/
- Click
Ok
- Click the green ✔️that is now visible
- Click the
+
sign
- Click the dropdown that says
- Click on the
-
There are many other configuration options for
sonarr
that are not covered here.sonarr
's webpage is here
variable | description |
---|---|
RADARR_PORT | the port that radarr will listen on |
<server-ip>:7878
radarr
should be configurable the same as any other installation of it. Feel free to skip these steps if you know how to configure radarr
already.
transmission
instead of the IP address when configuring the download client, as well as jackett
instead of the IP when setting up your indexer. This is because this uses docker-compose networking which means each service is accessible at the name of the service, rather than the host or even local IP address.
Step-by-step for those who need it
-
Add an Indexer
- Click on the
Settings
button at the top - Click on the
Indexers
tab - Click the big
+
symbol - Click the
Custom
button in theTorznab
section - Configure your
Torznab
feedName
: the name of this indexer (doesn't matter, just name it the name of your tracker)URL
: theCopy Torznab Feed
url fromjackett
that you saved earlier⚠️ It is necessary to replace the IP withjackett
http://jackett:9117/api/v2.0/indexers/<indexer>/results/torznab/
instead ofhttp://192.168.1.11:9117/api/v2.0/indexers/<indexer>/results/torznab/
API Key
: theAPI Key
fromJackett
that you saved earlier
- Click
Test
to verify that it is configured properly - Click
Save
- Click on the
-
Add a Download Client
- Click on the
Settings
button at the top - Click on the
Download Client
tab - Under
Completed Download Handling
toggleEnable
from Yes to No- We'll be using
filebot
to handle our completed downloads
- We'll be using
- Click the
Save
button at the top right - Click the large
+
button - Click on
transmission
- Configure your Download Client
Name
: whatever you want; probablytransmission
Host
:transmission
Port
:5656
or whatever you have set forTRANS_WEBUI_PORT
in your.env
fileUsername
:admin
or whatever you have set forTRANS_WEBUI_USER
in your.env
filePassword
:adminadmin
or whatever you have set forTRANS_WEBUI_PASS
in your.env
fileCategory
:radarr
- this allows transmission GC to properly remove torrents after seed limits exceeds
- Click
Test
to verify that it is configured properly - Click
Save
- Click on the
-
Add Some Movies
- Click on the
Add Movies
button at the top - Start typing in the search bar. It will search automatically when you stop typing
- Configure the download path (you should only have to do this on the first movie you add)
- Click the dropdown that says
Select Path
- Click
Add a different path
- Click the 📁button on the right of the modal
- Click
movies
# final path should be/movies/
- Click
Ok
- Click the green ✔️that is now visible
- Click the
+
sign
- Click the dropdown that says
- Click on the
-
There are many other configuration options for
radarr
that are not covered here.radarr
's webpage is here
variable | description |
---|---|
BAZARR_PORT | the port that bazar will listen on |
<server-ip>:6767
bazarr
should be configurable the same as any other installation of it. Feel free to skip these steps if you know how to configure bazarr
already.
Step-by-step for those who need it
-
Configure connection settings for Sonarr:
Hostname or IP Address
:sonarr
Listening Port
:8989
or whatever you have set forSONARR_PORT
in your.env
fileAPI Key
: theAPI Key
fromSonarr
-
Configure connection settings for Radarr:
Hostname or IP Address
:radarr
Listening Port
:7878
or whatever you have set forRADARR_PORT
in your.env
fileAPI Key
: theAPI Key
fromRadarr
<server-ip>:7676
variable | description |
---|---|
FILEBOT_PORT | the web interface port for the filebot container |
FILEBOT_FORMAT | the filebot format expression to use |
FILEBOT_ACTION | the action for filebot to take when renaming files |
FILEBOT_CONFLICT | what filebot does when it sees a conflicting filename |
FILEBOT_SERIES_DB | the database to use for TV metadata lookup |
FILEBOT_ANIME_DB | the database to use for anime metadata lookup |
FILEBOT_MOVIE_DB | the database to use for movie metadata lookup |
FILEBOT_MUSIC_DB | the database to use for music metadata lookup |
OPEN_SUB_USER | the opensubtitles username for filebot to use when downloading subtitles [NOT FUNCTIONAL] |
OPEN_SUB_PASS | the opensubtitles password for filebot to use when downloading subtitles [NOT FUNCTIONAL] |
The opensubtitles configuration is called by the container startup script, but opensubtitles still fails when filebot is actually run. Until I figure out how to make this work, stick to bazarr
.
You can navigate to <server-ip>:7676
to see the filebot command that will be run when a download is completed. Do NOT expose this port to the internet as it is not password protected.
These are the filebot
CLI options: https://www.filebot.net/cli.html. If there are additional options you want to be able to configure, open an issue.
4.9.x
version of filebot
by default. In order to properly register after you have purchased a license, copy your license.psm
file to the filebot config directory on your host machine. The container will automatically register filebot
with that license.
variable | description |
---|---|
TRANS_WEBUI_USER | the username with which to log into transmission |
TRANS_WEBUI_PASS | the password with which to log into transmission |
TRANS_WEBUI_PORT | the port for transmission's web interface |
TRANS_CONNECTION_PORT | the connection port for tranmission to use |
TRANS_MAX_RETENTION | the time in seconds before a torrent is automatically removed |
TRANS_MAX_RATIO | the ratio at which a torrent is automatically removed |
<server-ip>:5656
It should not be necessary to configure transmission beyond the default configuration. The container writes a config with reasonable defaults. If you need access to additional transmission settings, feel free to open an issue.
The container is already automatically configured to call filebot
to post-process a download.
<server-ip:7890
variable | description |
---|---|
NZBGET_PORT | the web interface port for nzbget |
The container is already automatically configured to call filebot
to post-process a download.
These are the automatic configurations in nzbget
now:
- The
MainDir
will automatically be set to/downloads/nzb
- The
ScriptDir
will automatically be set to/usr/local/bin
- The
Extensions
will automatically be set tonzbget-postprocess.sh
- The
ControlUsername
will automatically be set to whatever is configured inNZBGET_WEB_USER
iff it is set - The
ControlPassword
will automatically be set to whatever is configured inNZBGET_WEB_PASS
iff it is set
The paths are configured that way to support calling filebot
with the post-process script.
See the documentation for instructions on setting up the rest of nzbget
.
The first time that you set up your plex server, you will need to claim the server to associate it with your plex account. You need to access the server via localhost
or 127.0.0.1
in order to claim it. The easiest way to accomplish this is to create an SSH tunnel to your server so you can access plex on port 32400 on localhost.
ssh <server-ip> -L 32400:localhost:32400
Once you do this, you can now log into plex via localhost:32400/web
or 127.0.0.1:32400/web
and claim the server. Once the server has been claimed, you can log into it directly via the server's IP address.
variable | description |
---|---|
PLUGIN_LIST | a list of plugins to install. supported plugins are trakt and subzero . leave empty to install no plugins |
PLEX_WEB_PORT | the port for the plex web interface |
<server-ip>:32400/web
- Add some libraries
- TV Shows will be at
/data/TV Shows
assuming you followed the/media/TV Shows
convention forTV_DIR
- Movies will be at
/data/Movies
assuming you followed the/media/Movies
convention forMOVIES_DIR
- TV Shows will be at
⚠️ If you don't useBazarr
set up your media agents to not use local files (hopefully this will be fixed in the future)- There is a problem that I have not been able to fix yet where local TV Series art is not available. It appears the artwork downloaded by
filebot
is not readable byplex
for an unkown reason (I don't believe it's permissions related, but if you have ideas, please open an issue). - To get around this, uncheck
Local Media Assets
for all Agents underSettings
>Server
>Agents
. Artwork will be downloaded by plex and accessible.
- There is a problem that I have not been able to fix yet where local TV Series art is not available. It appears the artwork downloaded by
- Kill and restart the containers after logging in to Plex if you have Plexpass
docker stop plex
docker-compose up -d plex
The compose file includes a letsencrypt container which you can use to set up a reverse proxy to various services which will automatically provision an SSL certificate for you to use. If you do not want to use a reverse proxy, simply delete that entry from the compose file.
I recommend that you launch the full mediaserver once without the reverse proxy enabled so you can set up authentication for sonarr
, radarr
, bazarr
, nzbget
, tautulli
, netdata
, and heimdall
The reverse proxy is configured to use subdomain routing by default. It copies in appropriate configurations for each service with these names
service | config name |
---|---|
sonarr | s.subdomain.conf |
radarr | r.subdomain.conf |
transmission | t.subdomain.conf |
nzbget | n.subdomain.conf |
plex | p.subdomain.conf |
tautulli | u.subdomain.conf |
netdata | m.subdomain.conf |
heimdall | h.subdomain.conf |
bazarr | b.subdomain.conf |
Config files for each subdomain will be present in the config
directory on your host machine if you wish to change the configurations. The directory is config/letsencrypt/nginx/proxy-confs
. If you wish to use different subdomains (e.g. plex.domain.tld
instead of p.domain.tld
) you need to change the configuration for the subdomain in that directory, and update the LE_SUBDOMAINS
to include the new subdomain.
Note that the first time you run the letsencrypt container, it can take some time for it to register an SSL cert.
variable | description |
---|---|
LE_HOSTNAME | the hostname you are using to host the reverse proxy |
LE_EMAIL | the email for which your letsencrypt SSL cert will be registered |
LE_SUBDOMAINS | the subdomains for which to register the SSL cert |
I have built a wireguard container following the principles described here and have successfully tested using docker-compose to force all traffic for nzbget
through the wireguard VPN using a config from http://mullvad.net/.
Download or build a wireguard config file from your VPN provider. For example, mullvad has a wireguard config generator at https://mullvad.net/en/download/wireguard-config/. Once you have the config file (e.g. mine is called wgnet0.conf
, copy that file into your config/wireguard directory. The container will automatically find the config file and build the VPN network for you.
wireguard
config file. This will disallow traffic from the docker network which will make it inaccessible from the reverse proxy container. There is a rudimentary killswitch implemented in the wireguard
container already. I may improve this over time.
docker-compose.yml
has been updated with a commented out wireguard service. There is also a commented out nzbget
service configured to use the wireguard container's network as an example. This will force nzbget
to proxy all traffic through the wireguard
container, and thus through the VPN. You can follow the same pattern with the transmission
container, or even sonarr
and radarr
if you want all queries to torrent trackers to go through the VPN.
To proxy an additional container through the wireguard
container's network, add the service name (e.g. transmission
) to the list of network aliases defined in the wireguard
service.
networks:
default:
aliases:
- nzbget
- transmission
Then update the new service's definition (e.g. transmission
) to remove the port mapping list, and add
network_mode: "service:wireguard"
depends_on:
- wireguard
Because the wireguard
has an alias to transmission
, it is now accessible on transmission:5656
just like it was before. You will also want to add the port mapping that were originally in the transmission
service definition to the wireguard
service definition.
The wireguard
container now has a proper route back to the local network over eth0
so services that are using the wireguard
network should now be accessible via your local subnet. You need to specify this subnet in the LOCAL_NETWORK
environment variable. Thanks to htilly and cmulk who's containers I shamelessly copied and modified. Once you have set up the route back to the local subnet, you can properly port foward through the VPN. See https://mullvad.net/en/help/port-forwarding-and-mullvad/ for details on port forwarding with mullvad
.
Mounting docker.sock
is normally an anti-pattern for containers. However, when the wireguard
container is started, wg-quick
rewrites /etc/resolv.conf
with the DNS address specified in the interface's .conf
file. This breaks DNS resolution for docker services from within any container that is using the wireguard
container as its network. In practice, this means that from within e.g. the transmission
or nzbget
containers, curl filebot:7676
will fail DNS resolution. This breaks the auto-config between downlaoders and filebot
.
To work around this issue, the wireguard
container now inspects the network (docker network inspect mediaserver-docker_default
) and writes /etc/hosts
entries for each service it finds in the network. This means that if one of your service IPs changes for any reason, you'll need to restart the wireguard
container in order to be able to properly resolve DNS for docker services again.
I believe it is possible to fix this issue using dnsmasq
instead of writing /etc/hosts
which would remove the requirement to read from docker.sock
, but I have not been able to get it to work, so I'm leaving this hack in place for now.
If you do not mount docker.sock
, the wireguard
container will still run, but any containers in the wireguard
network will be unable to resolve DNS for other docker services. If that's fine with you, you don't need to mount docker.sock
.
The wireguard
container has been updated to use dnsmasq
to conditionally route DNS requests.
The container also now scrapes the wireguard interface file for its DNS server and write /etc/dnsmasq.conf with that as the final fallback DNS address.
There are now three options for enabling docker DNS resolution from within the wireguard container:
-
If
docker.sock
is mounted, write/etc/hosts
with any service in the network (this is really just for backwards comptability and is the same behavior as described above.) -
If
LOCAL_TLD
is set (e.g. to local) write/etc/dnsmasq.conf
to use127.0.0.11
for that TLD. Also, write/etc/resolv.conf
to searchLOCAL_TLD
so that containers can access the addresses of the services without having to know to append.local
to match the rule in/etc/dnsmasq.conf
. This will require aliases with the TLD in each of the containers that need to be accessible from within the wireguard network. An alias should look something like this (e.g. forLOCAL_TLD=ghost
).filebot: image: ghostserverd/filebot:4.9.x container_name: filebot restart: always networks: default: aliases: - filebot.ghost
which will result in
filebot
being accessible from containers within thewireguard
network. -
If
SERVICE_NAMES
is set (a list of services to make available from within the wireguard network), write each service name individually to/etc/dnsmasq.conf
to force127.0.0.11
as the DNS server for each service address. This is nice because you don't have to write an alias for each service to make available, but you do need to list out all of the services. A sampleSERVICE_NAMES
variable is set in thedocker-compose.yml
file. It shouldn't need to be modified, but if there is a reason to, open an issue and I can make it pull from the.env
file.
The last three settings are mutually exclusive. Only one of the mechanisms should be used. Of the last two, I'm not really sure which one I prefer, but they're both better than mounting docker.sock in my opinion.
Most of these containers are config wrappers around LinuxServer.io containers. Without their amazing linuxserver containers, none of this would have been possible. If you find this automated media server useful, go donate to them! They probably deserve it more than I do.
- forum.linuxserver.io
- IRC on freenode at
#linuxserver.io
- Podcast covers everything to do with getting the most from your Linux Server plus a focus on all things Docker and containerisation!
- Donate
This would also not be possible without filebot. This has been updated to use the latest 4.9.x
version of filebot. It supports automatic registration as long as you provide a license file.
Thanks to patorjk for his ascii text generator
Thanks to secretmapper
for the combustion
UI for transmission
If you really want to donate to me, you can do that here
- Consider switching from
nginx
totraefik
- Auto-configuration for linking
radarr
andsonarr
totransmission
- Auto-configuration for linking
radarr
andsonarr
tonzbget
- Auto-configuration for
plex
libraries - Improve documentation (maybe blog post with pictures)
Better configuration options fornzbget
Wireguard as a network container for downloadersAdditional containers (tautulli
,muximux
,portainer
)Add reverse proxy support (traefik
?)Upgradefilebot
to4.8.2
and make it easy to license