fabito / htk8s Goto Github PK

HTPC services running on Kubernetes

Makefile 26.25% Shell 73.75%

k3s kubernetes sonarr radarr bazarr emby-server htpc jackett argo-cd kustomize gitops emby transmission

htk8s's Introduction

HTPC powered by k3s

This is my current HTPC setup. It runs on k3s - a lightweight and easy to install Kubernetes distribution. It includes the following applications:

Sonarr for tv shows
Radarr for movies
Bazarr for subtitles
Transmission for torrents
Jackett for torrent tracker searching
Emby

Applications state (settings / db) and media files are stored in a shared volume of type hostPath. It does not use PVC and currently only works if the whole htpc namespace is deployed in the same node.

Getting Started

Quickstart

# for x86_64
kubectl apply -f https://raw.githubusercontent.com/fabito/htk8s/v0.1/install_x86_64.yaml

# for raspberry pi (ARM)
kubectl apply -f https://raw.githubusercontent.com/fabito/htk8s/v0.1/install_armhf.yaml

The Gitops way

# x86_64 only
kubectl apply -f https://raw.githubusercontent.com/fabito/htk8s/v0.1/install_argocd.yaml

This alternate manifest will install Argo CD along with the htpc application. Then it will monitor this repo for changes and apply them to the cluster accordingly (more specifically the overlays/x86overlay).

You can access the ArgoCD UI at: https://localhost/argocd.

Verifying the installation

All resources are created in the htpc namespace. So if you run:

k3s kubectl get all -n htpc

You should get something similar to:

NAME                                READY   STATUS    RESTARTS   AGE
pod/bazarr-795f88c5c9-w75l7         1/1     Running   0          24h
pod/emby-6f457df664-fqbmc           1/1     Running   0          24h
pod/jackett-6bcf6cd8d6-lrh6j        1/1     Running   0          24h
pod/radarr-5c965c7678-zt8sq         1/1     Running   0          24h
pod/sonarr-b65c8956-mxng4           1/1     Running   0          24h
pod/transmission-5f7fdc6cb5-nrtbb   1/1     Running   0          24h

NAME                   TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)    AGE
service/bazarr         ClusterIP   10.43.43.224    <none>        6767/TCP   24h
service/emby           ClusterIP   10.43.212.198   <none>        8096/TCP   24h
service/jackett        ClusterIP   10.43.104.233   <none>        9117/TCP   24h
service/radarr         ClusterIP   10.43.141.101   <none>        7878/TCP   24h
service/sonarr         ClusterIP   10.43.35.98     <none>        8989/TCP   24h
service/transmission   ClusterIP   10.43.184.198   <none>        9091/TCP   24h

NAME                           READY   UP-TO-DATE   AVAILABLE   AGE
deployment.apps/bazarr         1/1     1            1           24h
deployment.apps/emby           1/1     1            1           24h
deployment.apps/jackett        1/1     1            1           24h
deployment.apps/radarr         1/1     1            1           24h
deployment.apps/sonarr         1/1     1            1           24h
deployment.apps/transmission   1/1     1            1           24h

NAME                                      DESIRED   CURRENT   READY   AGE
replicaset.apps/bazarr-795f88c5c9         1         1         1       24h
replicaset.apps/emby-6f457df664           1         1         1       24h
replicaset.apps/jackett-6bcf6cd8d6        1         1         1       24h
replicaset.apps/radarr-5c965c7678         1         1         1       24h
replicaset.apps/sonarr-b65c8956           1         1         1       24h
replicaset.apps/transmission-5f7fdc6cb5   1         1         1       24h

You should also be able to reach each component's UI using the links below. Don't forget to replace localhost with the IP or the server name running k3s.

App	URI
radarr	http://localhost/radarr
sonarr	http://localhost/sonarr
bazarr	http://localhost/bazarr
jacket	http://localhost/jackett
transmission	http://localhost/transmission
emby	http://localhost/

Check the ingress-route.yaml for more details.

Each module except for Emby is configured to respond on a custom basepath (check the init containers logic for more details).

How it works (WIP)

It uses LinuxServers images.

It uses a hostPath volume to store configuration and media files. It defaults to the /opt/htpc directory

/opt/htpc
├── bazarr
├── downloads
├── emby
├── jackett
├── media
│   ├── movies
│   └── tv
├── radarr
├── sonarr
└── transmission

htk8s's People

Stargazers

Watchers

htk8s's Issues

Pre configure bazarr

Add init container logic to pre configure bazarr.

add bazarr to ingress
create /config/config/config.ini
configure sonarr
configure radarr
configure base_url

[assrt]
token = 

[opensubtitles]
username = 
use_tag_search = False
skip_wrong_fps = False
ssl = False
vip = False
timeout = 15
password = 

[sonarr]
apikey = abc
full_update = Daily
ip = sonarr
only_monitored = False
base_url = /sonarr
ssl = False
port = 8989

[proxy]
username = 
url = 
exclude = localhost,127.0.0.1
password = 
type = None
port = 

[radarr]
apikey = abc
full_update = Daily
ip = radarr
only_monitored = False
base_url = /radarr
ssl = False
port = 7878

[addic7ed]
username = 
password = 
random_agents = True

[legendastv]
username = 
password = 

[auth]
username = 
password = 
type = None

[general]
movie_default_hi = False
movie_default_language = []
ip = 0.0.0.0
use_scenename = True
use_postprocessing = False
enabled_providers = subscene,tvsubtitles
auto_update = True
port = 6767
use_radarr = True
base_url = /bazarr/
page_size = 25
minimum_score_movie = 70
branch = master
single_language = False
use_sonarr = True
serie_default_hi = False
path_mappings_movie = [['/movies', '/movies'], ['', ''], ['', ''], ['', ''], ['', '']]
serie_default_enabled = False
movie_default_enabled = False
serie_default_language = []
path_mappings = [['/tv', '/tv'], ['', ''], ['', ''], ['', ''], ['', '']]
postprocessing_cmd = 
minimum_score = 90
debug = False
use_embedded_subs = True
adaptive_searching = False

bazarr can not access from web

log from bazarr pod:
PermissionError: [Errno 13] Permission denied: '/config/bazarr.restart'

Maybe you need to add the chown part to the init container?

Duplicate value: "torrent-tcp" - how did you deal with kustomize bug?

kubernetes-sigs/kustomize#2934

after excute kustomize command I got duplicate port in install.yaml:

        - containerPort: 51413
          name: torrent-tcp
          protocol: TCP
        - containerPort: 51413
          name: torrent-tcp
          protocol: TCP

Volumes are not configured correctly for multi-node installation

The volumes are mounted per machine, instead of using PersistantVolumeClaim (PVC).
The issue is that In case more than one node is running, it's possible for example radarr to run on machine 1 and the download to be done on machine 2, meaning that radarr on machine 1 cannot copy the download from machine 2.

I think a starting 20-40gb is good enough.

Pending deprecation for Ingress extensions/v1beta1

Getting a deprecation warning when using a contemporary Kubernetes release. Just a matter of reformatting the spec for the Ingress resources. The deprecation was introduced in 1.14 and will remove it in 1.22+.

I'd be happy to submit a patch for it if you want, its a boring manual change. I am fixing it in my fork but it has a bunch of of other implementation specific changes.

If interested, how would you like this fix delivered? Separate fork or just a PR? Or do you not care (which is totally fair) :D

Default transmission configuration is damaged

Was checking this project out for the first time. Forgive me for my k8s/k3s laypersonship.

Ran the x86 install, was unable to access Transmission. Ingress said 502 Bad Gateway. Checking out the logs indicated that the settings.json was malformed. Inspection of the JSON shows:

"rpc-host-whitelist-enabled": false,

Obviously, something has been interpreted weird. So I scaled the replicaset to 0, nuked the file, and scaled back up:

kubectl scale --replicas=0 replicaset/transmission-5bf9655f77
rm /opt/htpc/transmission/settings.json
kubectl scale --replicas=1 replicaset/transmission-5bf9655f77

That restored it back to the application default which has the tragic side-effect of turning whitelisting back on. Back to the drawing board. Lets try a new deployment:

kubectl scale --replicas=0 replicaset/transmission-5bf9655f77
rm /opt/htpc/transmission/settings.json
kubectl rollout restart deployment/transmission

Isolated the issue to the deployment args by manually replacing the JSON being echo'd into the config file:

      - args:                                                                                                                                                                                          
        - echo starting; echo deadbeef > /config/settings.json; echo done;

Success. The config file literally just says "deadbeef." So there seems to be something wrong with the way the JSON is being spit out into the config file as a deployment argument.

Running on a bare-metal Ubuntu 20.04 server, if that matters.

VPN support

HI,

This setup looks great. I am thinking of trying it out shortly once I get a new pi 4. Any support for VPN inside the cluster ?