k8s-dev

Getting started

Storage

This example requires an LVM volume group named k8s-storage to be created before running the install. There should be at least 100GBs of available storage!

You can either create this Volume Group during installation, or to create it after the fact on a spare drive:

pvcreate /dev/sdb
vgcreate k8s-storage /dev/sdb

You should be able to run vgs and see the k8s-storage volume group as available -

# vgs
  VG          #PV #LV #SN Attr   VSize    VFree   
  k8s-storage   1   0   0 wz--n- <100.00g <100.00g

Network Configuration

k8s nodes do not like it when they change IP addresses. It's best to create a static DHCP reservation or use static IPs when deploying k8s.

Manifest deployments

Manifest deployments utilize kubectl and a group of manifest files to deploy services.

Deploy Entire Service

kubectl apply -f manifests/jellyfin

Deploy individual manifests

kubectl apply -f manifests/jellyfin/namespace.yaml

Deploy ALL services at once

kubectl apply --recursive -f manifests

To Remove Manifest Deployments

You can use the same syntax but with delete instead of apply. i.e. -

kubectl delete -f manifests/jellyfin

Or deleting the namespace for a deployment will remove all the resources contained within.

kubectl delete namespace jellyfin

Helm deployments

Helm is a tool used to simply deployments. You can use Helm 'Charts', which automatically handle the resources required for a deployment.

You may need to modify the default behavior of a Helm Chart. To do so, you'll use a 'values.yaml` file.

I've created bash scripts with examples of Helm deployments. Check out helm/zabbix for a good example.

cd helm/zabbix
./install.sh

Deployed Resources

This repo automatically sets up Pi-Hole running within the cluster. It also utilizes External-DNS to add local records to the Pi-Hole deployment. As such, you can point your laptop's DNS nameservers to your k8s-dev server to automatically enable DNS names for the following resources:

Service	Desc	DNS Hostname	Creds
garage-s3	S3 Storage Provider	s3.local	Auto-Generated
grafana	Grafana Dashboard	grafana.local	admin/admin
hastebin	Web Text Editor/Pastebin Alt	hastebin.local	N/A
jellyfin	Media Server	jellyfin.local	Created during first visit
nextcloud	Cloud Storage	nextcloud.local	admin/password (Defined in deployment.yaml)
pihole	Pi-Hole Admin Dashboard	pihole.local	password
plex	Media Server	plex.local	N/A
zabbix	Monitoring Service	zabbix.local	Admin/Zabbix

We also deploy a Wireguard Operator which dynamically allocates Load Balancer ports.

Cronjobs

We install a couple cronjobs to show examples of how they work. They can be found in manifests/cronjob

Here's an example of viewing their output -

root@k8s-dev:~/k8s-dev/manifests/cronjobs# kubectl get cronjob
NAME           SCHEDULE      TIMEZONE   SUSPEND   ACTIVE   LAST SCHEDULE   AGE
curl-an-api    */5 * * * *   <none>     False     2        65s             8m15s
do-some-math   * * * * *     <none>     False     0        5s              28m
root@k8s-dev:~/k8s-dev/manifests/cronjobs# kubectl get pods 
NAME                          READY   STATUS             RESTARTS      AGE
curl-an-api-29451211-2c5lc    0/1     CrashLoopBackOff   5 (2m ago)    5m10s
curl-an-api-29451212-72995    0/1     CrashLoopBackOff   5 (78s ago)   4m10s
curl-an-api-29451215-5w94w    0/1     Completed          0             70s
do-some-math-29451214-dbb6k   0/1     Completed          0             2m10s
do-some-math-29451215-rvjtj   0/1     Completed          0             70s
do-some-math-29451216-bjqqw   0/1     Completed          0             10s
root@k8s-dev:~/k8s-dev/manifests/cronjobs# kubectl logs curl-an-api-29451215-5w94w
{"latitude":52.52,"longitude":13.419998,"generationtime_ms":0.024199485778808594,"utc_offset_seconds":0,"timezone":"GMT","timezone_abbreviation":"GMT","elevation":38.0,"current_units":{"time":"iso8601","interval":"seconds","temperature_2m":"°F"},"current":{"time":"2025-12-30T05:30","interval":900,"temperature_2m":30.7}}