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.
Should I be seeing other servers on the same "LAN" as my VPS?
voltagex@frankfurt:~$ avahi-browse -a
+ eth0 IPv4 TorrServer Web Site local
+ eth0 IPv4 TorrServer _torrserver._tcp local
+ eth0 IPv6 TorrServer _torrserver._tcp local
+ eth0 IPv6 TorrServer Web Site local
+ eth0 IPv6 Home Assistant _home-assistant._tcp local
+ eth0 IPv4 Bubby (OpenClaw) (2) _openclaw-gw._tcp local
+ eth0 IPv4 openclaw-dr _openclaw-gw._tcp local
+ eth0 IPv4 DE20USD Microsoft Windows Network local
+ eth0 IPv4 DE20USD Device Info local
Comments
Did you create them yourself?
No, these aren't mine.
Srsly? OMG. Who's the provider?
Yeah who's the provider? Name and shame
I'm not putting them on blast as I'm waiting for tickets to be resolved and working on locking down my server.
Initial response is good-ish.
Dunno what the rules are about that here.
just wanna know the provider so we can avoid any issues with our own VPSes. especially with some of the things we run on our VPSes.
Will give them a chance to change settings on their end first.
They have suggested turning on ufw.
The issue is far greater than a single guest enabling a firewall...
Then don't open a thread if you don't want to name the provider. Rules here allows this
I guess it MassiveGrid
OP's only comment that involves some form of hosting provider was with MassiveGrid (Nov 2024)
LOL LMAO
I don’t think it’s just one provider. Tested the below and all of them have entries like this.
DartNode, Hostsailor, Hostdzire, VPSLayer, AspireHost
LowEndDetectives
LMAO WTF
oh wow... gotta check this on mine now
God damnit. Add SolidVPS to the list. My guess is that we should be looking for VPS that have this configured correctly because the default is going to be shared layered 2 broadcast.
@yoursunny time for another list
Every provider I've tried is showing at least one other host. It seems wrong, but I'll leave it to other more knowledgeable people to say whether this is just something that is industry standard or not. It seems you can mitigate by shutting down the avahi daemon (and ostensibly anything else that broadcasts mDNS) and enabling a firewall rule that blocks inbound 5353. So if you're running the default ufw block pretty much everything, I think you're fine.
Many providers will have this. They generally put you in the same /24 and (hopefully) filter from there. It shouldn't be a big deal if they have ARP spoofing protections and such in place.
I found it was easiest/safest to have a fully routed infrastructure.
Wonder what happens if you define all the ipv4 of the subnet on your VM.
It won't work obviously, everyone has anti spoofing setup(Unless some provider doesn't).
but they are on localhost, screw spoofing lot of basic vps users will lock external access to local only.
This could be used to bypass limits, and abuse exploits.
In the age of the vibe coded panel, hosting company, support and potentially owner, I don't know if we can be sure of anything ;-)
Hi,
i just had today a ticket like this. This was my answer:
And that still stands, and i am wondering a bit about this shocking moment of some people here.
I do not see much difference if you are using a software to tell you what is available in your broadcast domain or if you are using a port scanner.
For one of the monitoring nodes at hetzner i just did that:
With here some selected examples:
So what exactly is the difference now?
Except the fact that because we configure vanilla IPs without the required on-link config directive and being this way also compatible to older OS's some customers might want/need.
Yes, this is expected behavior.
When you share the same /24 subnet with another server, your ARP and other broadcasts must reach each other according to the Internet Protocol.
If the switch blocks such traffic, you’ll be unable to properly communicate with another server in the same subnet.
There are legitimate use cases for such communication, such as (1) you bought two servers, one as frontend, the other as database (2) you want to fetch from a website hosted by someone else who happened to have chosen the same provider and deployed in the same subnet.
Crazy amount of lack of competence on display here.
avahi-browsegives you services discovered through mDNS, which is based on multicast. It has nothing to do with subnet adjacencies / neighbors. Reaching open services within the same subnet works regardless of whether the virtual machines share a broadcast domain or not - it's entirely irrelevant.Allowing BUM traffic between virtual machines indicates that the provider doesn't have a production-ready guest filtering setup. At least they have uRPF and don't allow VMs to send traffic from random IPs - right..., right...?
To answer OP: on any serious provider's network the answer is nope.
Should this be read as "not a vibe coded integration between a base proxmox installation and a dodgy panel hacked together by chatGPT free tier"?
The LAN should go through but the mDNS shouldnt.
You should "see" the other neighbors as in your ping can reach them (provided not blocked by firewall) or as in you can reach their website hosted on that IP.
...But you shouldn't be able to discover them via
avahi-browse.