docker-htpc

:warning: This readme is NOT up to date. Just use the pieces as an example for your own configs.

Containers for HTPC apps.

All containers are built from the ubuntu:trusty base image because most of the apps publish official .deb distributions for Ubuntu.

WARNING: This is a work in progress and suits my own specific needs. Feel free
to use pieces of it but I wouldn't recommend using it wholesale unless your
HTPC server setup is magically identical to mine.

init

All containers use s6-overlay for process supervision and pre-startup scripts to fix permissions on the /config volumes.

user

The apps inside the containers each run as the nobody user for security. Permission dropping is handled by s6-overlay. Also, s6-overlay will run chown -R nobody /config inside each container during startup to fix up the perms in the /config volumes.

Volumes

Each container generally follows a similar pattern for config and data volumes:

• /files -> /files: Large data volume. In my case, there are folders such as: /files/tv_shows, /files/movies, /files/downloads, etc.

• /etc/<container_name> -> /config: eg: /etc/sabnzbd/ on the host is mapped to /config inside the sabnzbd container for persistent config storage.

This is not the case for all containers. Check the Makefile for details on each container.

TODO: enumerate each container's host volume mounts here.

Ports

• organizr:
  • 8282: http web ui
• muximux:
  • 8080: http web ui
• tautulli:
  • 8181: http web ui
• sabzbd:
  • 8085: http web ui
• sonarr:
  • 8989: http web ui
• deluge:
  • 8083: http web ui
  • Note: the deluge container is run with --net=host in order to allow deluge to punch holes with NAT-PMP. It will work fine without --net=host however, perhaps with limited connectivity to some torrent peers but more secure.
• plex:
  • network: host
• embyserver:
  • network: host
• timecapsule:
  • network: macvlan.
    • This container will be allocated its own IP on your local network based on the macvlan config.
    • Use docker inspect to view the allocated IP
    • See timecapsule (Samba) for more details.
• unifi:
  • network: macvlan.
    • This container will be allocated its own IP on your local network based on the macvlan config.
    • Use docker inspect to view the allocated IP
• grafana:
  • 3000: http web ui
• graphite-api:
  • network: metrics
  • No forwarded ports
• go-carbon:
  • 2003: graphite line protocol
• influxdb:
  • network: metrics
  • No forwarded ports
• telegraf-snmp-unifi:
  • network: metrics
  • No forwarded ports

Note: most of these apps can also expose TLS https ports but the current config does not expose these.

Usage

List tasks and descriptions

$ make help

Create and start containers

$ make start

Containers will be created, started, and set with --restart=always flag.

Update (rebuild + recreate) all containers

$ make rebuild-all

Update (rebuild + recreate) specific containers

$ make rebuild CONTAINER=<name>

Start/Stop/Restart individual containers

No make tasks for these since they're simple docker commands:

$ docker ps
$ docker stop sabnzbd
$ docker start sabnzbd
$ docker restart sabnzbd

View container logs

Use docker logs command:

$ docker logs sonarr
$ docker logs -f sonarr

Notes

deluge autoconnect

Sonarr will not be able to connect to deluge unless you login to the deluge-web client and select the server. This needs to be done once per startup of the deluge container, unless auto-connect is enabled. Follow these directions to setup deluge-web auto-connect: http://dev.deluge-torrent.org/wiki/Faq#HowdoIauto-connecttoaspecificdaemon

timecapsule (Samba)

This container expects to connect to a docker network named localnet. This network should be created with the macvlan or ipvlan drivers available in docker 1.12+. This allows the container to be started and acquire its own IP on your local network in order to run Bonjour (avahi) services to announce itself on the network.

This is particular to my setup because I have an existing samba instance running on the host providing general file sharing services. This container acts as a separate device that looks roughly like a timecapsule on the network exporting only a single timemachine mount.

It is probably possible to use standard Docker port mapping instead. See the udp and tcp ports exposed in timecapsule/Dockerfile. However, getting the avahi announcement to work could be challenging. A workaround would be to publish the timecapsule services from the host's avahi using the timecapsule/smb.service file.

If your local network is 192.168.0.0/24 you can create a localnet network with the following docker command:

$ docker network create -d macvlan \
    --subnet 192.168.0.0/24 \
    --gateway 192.168.0.1 \
    -o parent=eth0 \
    localnet

TimeMachine will attempt to use the entire disk by default. We constrain it by lying about the max available space on the samba share. This is done using a dfree script. By default, the max space presented to TimeMachine is 750GB. This can be changed by updating the TIMEMACHINE_MAX_VOL_SIZE_GB environment variable in the Dockerfile.

metrics (grafana, graphite-api, go-carbon)

All three metrics containers are attached to a user-defined bridge network named metrics. The only forwarded port is port 3000 (http) to the grafana host.

Metrics are stored in /etc/go-carbon/data which is mounted into both the go-carbon and graphite-api containers.

grafana fetches metrics from the graphite-api container over port 8000.

Metrics are sent to port 2003 which is forwarded into the go-carbon container, example from host:

    echo "testdata.foo.bar $RANDOM $(date +%s)" | tee >&2 | nc localhost 2003