Had some big outage because of… Scaleway ?
The egress for my K3S cluster is hosted there and was for some reason unavailable
I will have to track the logs, but not being able to login the main console to hard reboot the instance didn’t help
Time to bring that multi ingress dns rotating k8s operator project out of the forgejo limbo
#homelab #selfhosted #selfhosting #k3s #k8s #kubernetes #cloud #scaleway #outage #dns #ingress #egree
@whyhankee
A quite robust solution.
With its choice of the MicroOS (immutable server distro from openSuSE) as a host OS and a bit of configuration efforts of system-upgrade-controller and kured to automate the host and the kubernetes upgrades you may get a cluster that keep itself healthy and actual with relatively low effort.
I'd say it's easier to maintain than a AWS EKS one.
Caveats:
- learn Kubernetes concepts and cli management
Recommendations:
- hetzner k3s terraform config offers node groups, use small ARM instances for the control plane
- use S3(-like) storage to backup etcd (and Rancher) state, thus making your cluster recreatable from offsite state
- heterogenous cluster works and is cheaper, if your workload allows it
- enable Rancher or install it yourself, it's way better than UI in AWS/GCP
Feel free to ask me about it if you have some particular question.
It's time to learn about Kubernetes.
I know a tiny bit about K3S and managed to get a cluster of two machines running via Ansible Playbook, but I'm not familiar with it whatsoever.
However, I know this is an important tool and I need to learn how to use it so that's exactly what I'm gonna do.
If any of you can recommend tutorials or sites teaching it, I'd appreciate you writing it down in the comments.
#ansible #playbook #deployment #k3s #kubernetes #cluster #learning #sysadmin
I can finally receive, send and reply to and with emails of my own domain(s)
That took 7 years of self hosting, a lot of research and, admittedly, a SMTP relay for outgoing emails, because the extra work to maintain a proper IP reputation (among others) didn’t seem worth it (yet ?)
Using Stalwart helped, being on K3S didn’t…
All that’s left if properly hooking up Postgres, some storage engine and a k8s configmap to track the Stalwart configuration file
#selfhosted #selfhosting #kubernetes #homelab #k3s #networking #k8s #email #mail #smtp #stalwart #postgres
Nu e heimelaben min: to ~10år gamle stasjonær-PCa me #nixOS, tilkobla UPS, én gammel laptop me Bazzite, og én vps nedi oslo me nixOS.
De kjøra alle kubernetes i form av #k3s, som holdes i tøylan av #fluxcd.
nixOS tøyleholdes av #comin.
Kikk gjerne: https://codeberg.org/papiris/infra-cd og /nix-config
Neste bi å få opp monitorering så æ kan hold beire øye me krætturet, og offsite backup av nåkka slag.
Rop ut om du vil bi me på å lag ei gjensidig lagringsklynga basert på sjøldrifta S3: https://garagehq.deuxfleurs.fr
A little bit of #Kubernetes al a #K3s to start a Saturday morning. Getting ready to down a node for installation of a GPU for hosted LLM workloads. What is the weekend #homelab project for you this week?
While I was away, I finally got NetBird back up and running, and damn it was hard but it’s worth the effort
Setting it up with PocketID for authentication was probably the most headachy thing I could’ve pulled off, but now that it’s done and stable, I ain’t touching it
If you are looking for a stable and fully self hosted Tailscale replacement, I highly recommend !
#tailscale #headscale #netbird #vpn #wireguard #selfhosted #selfhosting #homelab #pocketid #kubernetes #k3s
Mein #freudiverse der Woche kommt dieses mal von meiner Arbeit.
Ich habe die Webseite für einen neuen Betriebszweig unseres Demter Hofes fertig gestellt: Schnittblumen lokal und nachhaltig angebaut.
Vor allem freue ich mich für meine Freundin, die Ihr Herzensprojekt jetzt umsetzen kann. Und die #bienen werde es lieben🐝
Für die Technik Nerds: Die Seite basiert auf #odoo Community Edition. #k3s #ansible #hetzner
In der Firma haben wir unsere #matrixorg von #Synapse auf ESS (#k3s) pro migrieren lassen. Das hat erstaunlich gut geklappt. Auch alte LDAP Accounts sind auf #oidc (Azure) gemappt worden. Es gab nur wenige Probleme, die aber hauptsächlich am #element lagen und der vielen Bugs.
Meinen privaten Server habe ich auf die Community Version umgebogen, allerdings hat bei mir die Migration nicht geklappt. Nicht dramatisch. Die alte DB war eh noch voll von Spam. Was noch nicht klappt: Element Call von Privat zur Firma. Aber von Firma zu Privat klappt. Allerdings hängt mein Server hinter einer #Haproxy hab noch keine Ahnung was das Problem ist.
So I saw a blog post linked on here the other week about someone's homelab where they use #Incus to run all their containers and VMs. And I've fallen into the rabbit hole.
That post: https://linderud.dev/blog/personal-infrastructure-setup-2026/
Anyway, Incus is the fork/successor to #LXD which as recently as last year they released #IncusOS which is a very slimmed down OS for running multiple types of containers and VMs. A bit like #Proxmox in a sense.
What I like about the distro:
- Immutable
- A-side/B-side partition layout for friendlier updates
- Requires Secure Boot + TPM, resulting in encrypted drives by default
- ZFS. I've loved ZFS for many years.
It really seems like this was built for edge type deployments where secure "appliance" like things really excel yet still a net benefit elsewhere.
Since vSphere was killed, this feels pretty damn close to what I liked about it.
Can already run VMs along side "System Containers" (shared kernel + init system) and "App Containers" (what everyone calls "Docker"). I see on the roadmap support for MicroVMs (OCI container + individual kernel).
I run all of my stuff bar storage on #k3s on baremetal but there's times when I need a VM or different container behaviour than it offers.
[Show GN: EdgeFlow - 수평 확장 가능한 실시간 영상 분산 처리 프레임워크
EdgeFlow는 라즈베리파이와 같은 저성능 기기에서 고fps 영상 처리를 가능하게 하는 k3s 기반의 분산 처리 프레임워크입니다. Redis를 통해 영상 프레임을 여러 처리 노드로 분산하고 다시 취합하는 방식으로 수평 확장을 구현하며, YOLOv5의 성능을 선형적으로 향상시킬 수 있습니다. 주요 특징으로는 직관적인 파이프라인 구성, 유연한 관리, 프레임 순서 보장, 로컬 테스트 등이 있습니다.
https://news.hada.io/topic?id=26349
#edgeflow #distributedprocessing #yolov5 #k3s #realtimevideo
We've had a number of people interested in this while we were walking around at #FOSDEM so let's give some insights here..
We have 3 physical machines in an #EU located and owned datacenter. We run a #Ceph storage cluster and #K3S on top of them.
Our operation is fully #gitops using #ArgoCD. We run our site with #offen for basic analytics, so visitors have #privacy, #transparency and #control.
If you see someone in a black shirt and our logo during #fosdem2026, feel free to approach us!
Mal eben den RAM beim NUC upgraden…
Die Chromebox3, die andauernd neu startet hat RAM, die ein NUC gut gebrauchen kann.
Der Plan: NUC drainen, System-Updates einspielen, herunterfahren, RAM einbauen, hochfahren, fertig.
Realität:
NUC drainen, System-Updates einspielen, herunterfahren, RAM einbauen, feststellen: och, die NVME-SSD ist ja viel kleiner als die der Chromebox3 …
Also auch die SSD umbauen und das System „mal eben“ neu installieren.
— dann sagte das BIOS: CMOS Error. —> Die Batterie ist leer
Dafür hat man ja passende CR2032 zu Hause. Dumm nur, dass hier die Batterie per Molex-Stecker auf dem Mainboard sitzt.
An einer Batterie wollte ich auch nicht rumlöten.
Also sitzt der NUC jetzt im Wohnzimmer, wg. LAN und Monitor und spielt dort so lange bis die Austauschbatterien da sind.
Die Neuinstallation habe ich trotzdem gemacht und dann kam es zum "Waiting for other members to finish joining etcd cluster: etcdserver: unhealthy cluster“.
Also erst einmal den etcd vom k3s angeschaut... und da gab es einen (laufenden Node 2x) …
Was für eine Freude. Okay, es läuft nun alles wieder.
Die Lektion aus der Geschichte:
- seltener verwendete Kommandos wie „irgendwas mit etcdctl“ dokumentieren
- ab und zu einmal Geräte vom Strom trennen um defekte Batterien zu finden
Schluss mit Blindflug!
In Teil 6 der Rebel HomeBase bauen wir den ultimativen Wachturm auf dem ZimaBlade. Wir installieren Prometheus & Grafana für Transparenz im K3s-Cluster.
Wir zeigen nicht nur das Ergebnis sondern auch den Weg durch die YAML-Hölle und das Rechte-Drama. Echte IT, ungeschönt.
Wer seine Infrastruktur beherrscht, muss sie auch sehen können. ✊
https://www.pandolin.io/rebel-homebase-6-monitoring-grafana-zimablade
#RebelHomebase #Grafana #SelfHosted #K3s #Linux #OpenSource #Ubuntu #BeAVoice
https://www.pandolin.io/rebel-homebase-6-monitoring-grafana-zimablade/
Toujours un plaisir renouvelé d'obtenir ce résultat après reboot du pitit cluster k3s #selfhosting #k3s #argocd
I took a break from blogging about SimKube this week to instead talk about my adventures setting up ACRL's first ((semi-)permanent) Kubernetes cluster! I had fun writing this (and doing the work), so hopefully you have fun reading it:
https://blog.appliedcomputing.io/p/what-to-expect-when-youre-expecting
Das Gehirn der Rebellion ist online!
In Teil 5 der Rebel Homebase lassen wir die „Lego Duplo“-Welt der Hyperscaler hinter uns.
K3s auf Blech: 5 Nodes, 1 Mission. Pure ARM-Power statt Cloud-Miete.
Schluss mit Cloud-Illusionen und dem digitalen Stockholm-Syndrom. Wir bauen Infrastruktur, die uns gehört. ✊
https://www.pandolin.io/k3s-blade-cluster-aufbau-cm4-cm5/
#RebelHomebase #Kubernetes #K3s #OpenSource #BeAVoice #Linux #HomeLab #DigitaleSouveranitat #Ubuntu #CloudRebelAlliance
#k3s or #k3d? Is there even a difference - and also, what are they useful for? Is it really only good for a quick 'throw-away' #kubernetes cluster for testing, or something? coming from something like #rke2 (which i know is prolly not a good comparison, but still curious how they could be useful to me)
