New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.

Comments
Upgraded...
GRA seems to be where it's at for KS-2s, 4x4 TB squad
Also 10Gb down
I have a GRA KS-LE-C (E5-1630v4, 64 GB RAM, 2x480 GB SDD) which next renewal is 18 Dec 2025 available for transfer. If you want it, dm me. OVH CA World account is needed.
What is the renewal rate?
14.20 usd/month
isn't it beautiful?
premium LE-B Idle Cluster
you need more
Nice Chickens.
https://stats.neoon.net/dedi
I don't cluster mine, I want them to be strong and independent chickens.
Took me and gemini a few hours to get it clustered over vpn (just give me vrack ovh, the chickens are on the same physical rack....) with migration working, v6 routing, v4 nat, etc.
Not sure if I would recommend it, but I got so used to live migrations in my homelab pve cluster
Hi guys,
I´m wondering if anyone of you has an idling hybrid server willing to transfer to me? Preferably with 2x 4TB HDDs ...
It´s year end, time to clean up your server portfolios
P.S. I already have a pending order at @allthemtings
Cleanup? No, fill the portfolio before prices raise.
No, no, you need to make some room before the next Mystery sale 😂
Actually pretty easy and great with tailscale and SDN. Setup tailscale on every node, cluster using the cli with the tailscale IP on link0 and setup a EVPN/VXLAN SDN in the GUI. Bam, you got a L2/L3 network across the nodes with internet access on every VM. Depending on the configuration either using one uplink or every uplink in the cluster.
That's close to how I did it.
Didn't use EVPN tho.
They get the "normal" ipv6 treatment from the guide flowing around here with ndppd and ipv4 NAT
Slaac takes care of giving them new v6 after migration
Annoying that proxmox DHCP and ipam are not working with vxlan, but what can you do
Yeah, they wanted to add this feature since 2023 and it's on their devmail since. It's just too buggy as they say. At least it gave me a reason to deploy my own using AdGuard Home. It not being HA isn't even concerning for me as the leases last at least 24 hours in my configuration and most of the time I assign static IPs anyways or at least map MAC to IP on DHCP server for something like talos linux.
Hi,
I’m currently using a KS-STOR server and overall it’s a great machine.
However, I’m starting to need more RAM, so I’ve been looking at the SYS-STOR servers available in France (Roubaix).
I’d like to ask about your experience with network speeds on SYS-STOR:
Do they really have 10 Gbit download?
And how is the upload speed in practice, ideally also 10 Gbit?
Any real-world feedback would be greatly appreciated.
Thanks a lot!
In most times it is going to be 10G down, sometimes there would be 10G up too (KS only, SYS-STOR never seen before, but I guess you already have 1G up)
This mega thread shouldn't be on the second page, so let's go.
Real HA for applications (here beszel) by using ZeroTier + Tailscale + Proxmox + Talos on Kimsufi. Oh I will have my fun writing a detailed tutorial.
I use way too much time to make Kimsufi enterprise production ready, what do I do
How are you using ceph?
Do you use the proxmox ceph for k8s or the other way around? Different cephs? Cephception?
Why ZeroTier + Tailscale? They both do the same thing.
For a (fairly static) cluster you might as well use regular wireguard with a config assistant like
innernetto squeeze out a tiny bit more of performance.In this specific cluster I use the regular installation with xfs, leaving me more RAM than with zfs. The CEPH on the k8s is a rook-ceph that has one additonal mounted disk per k8s node, could use longhorn too, but rook-ceph is a bit more stable in my expirience.
Well, you would think so. I want the cluster to be easy to setup and maintain, while being able to use pretty much everything I want. Tailscale limits you to a MTU of 1280, if you increase the MTU, there will be problems with DERPs or even between devices. As I use EVPN-VXLAN SDN to simulate a L2 network across the Proxmox nodes, I need a (virtual) MTU of at least 1500 to support all applications. ZeroTier automatically uses a MTU of 2800 and does the heavy work of fragmentation itself.
ZeroTier: used as the underlying private network between the Proxmox nodes, to allow a 1500 (virtual) MTU on VMs.
Tailscale: used as a "access" VPN to allow safe access from outside and automatic failover access to the simulated subnet using multiple deployed subnet router.
KS-Stor is now unavailable in every region, perhaps in preparation to refresh it?
New whole KS-Line is on the way this or next month. I wouldn't expect them to refresh a single one, but with OVH, you never know.
Anyone interested in getting rid of any idlers they have?
+1, anyone looking for get rid of their idlers?
I think @TomasSystems was getting rid of a few KS-N's might be worth dropping him a message
Ah it's good you're posting, I totally forgot to thank you for the KS-A. I asked for a disk replacement, let see if they allow it.
Please forward all KS-A related questions to @Wolf who has TWO of them
Anyone interested in an KS-1 (now KS-2) with 4x2TB in BHS with an EU account?
Will be available in the next few days.