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
Yes, probably the reason as this is a transferred server. Thank you for confirming that theory.
But:
Having to contact OVH support is one of the most dreadful thoughts you have to live up to. But I will be brave and see what will happen this time.
Chat is the way I discovered
OVH support is absolutely atrocious and having to pay for a support plan is even more atrocious just to get help for something you are already paying for
What is it, you need?
Everything you would need normally is covered in the included support. They only provide the network infrastructure as well as the hardware, that's it.
I have one KS-LE-B with random restart, first they tested mem/cpu for 10+10min said its fine, but restarts are like 2-3x times per week so now i wait for their response as test defo can not detect those, this is 3th restart last week so hope they exchange psu and mobo
linux do not log any issue just plan reboot
Just tested this on my 2025 KS-LE-E box, the server does not respond to IPv6 ranges beyond the allocated /128 IP even after replacing the default /128 with /64 in the network configs as per instructions in:
https://help.ovhcloud.com/csm/en-dedicated-servers-network-ipv6?id=kb_article_view&sysparm_article=KB0043765
Also, https://github.com/ovh/infrastructure-roadmap/issues/240
They must have changed something 🙁
Can you verify that on any of your 2025 Kimsufis? I am not skilled in Linux networking.
I was able to use the /64 on all my KS servers (2024 LE-B, KS-LE-1, KS-LE 2025, KS-MYSTERY)
So double check your config
Which OS are you using? I updated /128 to /64 in the /etc/netplan/50-cloud-init.yaml on Debian 13 and rebooted. The server was only able to ping to a:b:c:d:0:0:0:1 but not to a:b:c:d:0:0:0:2 after the update. What am I missing?
I use debian 13 / nixOS / archlinux. But i'm not using netplan.
I'm pretty sure your config is the problem, don't have time to deep dive right now. But if you want to test you have the whole /64 assigned to your KS
just assign another address manually
2001:db8:abcd:0001::1/64
to
2001:db8:abcd:0001::a/64
If you want the /64 to reach your interface, there is probably something else to configure (ndp related probably)
You need to setup NDP with something like ndppd or anything else that get's the job done. It's a /56 that is assinged to nearly every server on OVH, which obviously allows to use the full /64 of the /128 address.
you need ndppd and ipv6 forwarding.
Are you sure it didn't reset after reboot through cloud-init and that's why there's the
/128again?You can use chattr +i FILE command to make the file immutable.
Strangely, changing the default a:b:c:d::1/128 to a:b:c:d::1/64 in the netplan yaml had no effect as a:b:c:d::2 was still not reachable.
But adding another entry a:b:c:d::2/128 to the same netplan yaml made that reachable. Yet specifying a /64 instead of /128 does NOT make the other IP addresses in that same subnet reachable even in case of the additional entry.
This is what Google Gemini said on the topic:
This is a classic configuration issue specific to hosting providers like OVH, Hetzner, and SoYouStart. In these environments, the IPv6 gateway (...ff:ff:ff:ff:ff) is technically "off-link"—it exists outside your assigned /64 subnet. Standard networking rules (and Netplan's default behavior) block you from talking to a gateway that isn't in your subnet. To fix this, you cannot simply combine the default route and the on-link flag. You must define two separate routes: A specific "link-scope" route to reach the Gateway itself. A default route that uses that Gateway. Additionally, simply setting the subnet to /64 does not make your server "listen" on all addresses in that subnet; it only creates the routing path. You must explicitly list every IP you want to use.A bunch of VM's just flatlined, bloody hell, 450GB NVMe isn't enough.
I think for Kimsufi the gateway doesn't route the whole prefix, but only individual IPs. It means that only the ::1 is routable by default and you need to explicitly advertise other addresses.
I use systemd-networkd and IPv6 works on my KS servers with this configuration:
Imagine configuring each IPv6 address, ugh.
Just use ndppd and you can use any IPv6 without configuration.
Haha, true. I only have a handful of static IPv6 though, so this works as a set and forget. For a bunch of VMs on a bridge interface ndppd would be the way to go.
some news serwers in API?
We know about these, we expect them for Black Friday.
understandable
New stock, yes please.
KS-LE-E rental - datacenter rbx - 2x SSD SATA 480GB
KS-LE-A rental - datacenter rbx - 3x 2TB HDD
I currently have 2 pending invoices for the servers above. Before I pay, I wanted to ask: Did anyone get any lucky specs or upgrades for these recently?
My paid invoice for a
KS-LE-A rental - datacenter rbx - 3x 2TB HDD
was cancelled last Friday ...
BF sales will most likely start this week, orders would be cancelled, I guess
They wouldnt do this to me while I'm away from home on a long weekend with only my phone and unstable (at best) mobile data. Octave wouldn't ? Would he?
Anyone eyeing up some of these for black friday?