Sometimes, Proxmox is not visible on the network, before I restart it a couple of times.

[yonoss]

Ever since I've discovered proxmox (about 6 month ago) I noticed a weird glitch that is only visible on proxmox system: After a power on or a restart, the proxmox server is not accessible/visible on the network. There are no problems though reported on the proxmox console when this is happening. From the console side it looks like everything is ok. But still, the system is not accessible from the network.

To overcome this problem I have to restart the proxmox insance a couple of time, and on its own it will become visible.

I have a router that provides DHCP functionality, and I can tell you that, this problem is only visible with the proxmox instances. Everything else works ok. There are no IP conflicts on the network either.

I noticed the problem on multiple proxmox instance (I've installed it more than 10 times on different systems, and on most of those setups I saw this problem). I've played with version 8 and version 8.1. So is definitely not an installation issue, from what I can tell.

Any idea what could be causing this behavior?

Thanks

 

[lildadou]

what does "system is not accessible from the network" mean? it doesn't respond to ARP requests? to pings? it refuses SSH connections?

did you at least check that your interfaces had an IP? (ip addr)

 

[yonoss]

when this is happening I cannot access the proxmox web console and I cannot ssh to the instance. And yes the system shows it has an ip associated.

 

[leesteken]

What is the output of cat /etc/network/interfaces? Is the static IP of Proxmox inside the DHCP range of the router?

 

[yonoss]

This is the output of the cat, when the system is accessible. When I'll catch it again in that weird state, I will rerun it.

root@proxmox1:~# cat /etc/network/interfaces
auto lo
iface lo inet loopback


iface eno1 inet manual


auto vmbr0
iface vmbr0 inet static
        address 192.168.1.231/24
        gateway 192.168.1.1
        bridge-ports eno1
        bridge-stp off
        bridge-fd 0


iface eth1 inet manual




source /etc/network/interfaces.d/*
root@proxmox1:~#

And yes the proxmox ip is in the dhcp range.
As I've mentioned the problem gets solved, without doing anything special, after a couple of restarts.

 

[leesteken]

This is the output of the cat, when the system is accessible. When I'll catch it again in that weird state, I will rerun it.

No need it is a static file and does not change.

address 192.168.1.231/24

And yes the proxmox ip is in the dhcp range.

If you fix that, the problem will probably go away. It's not uncommon that overlap between static and dynamics addresses causes duplicate addresses and problems like you describe.

 

[Nollimox]

If you fix that, the problem will probably go away.

Yes, Leesteken is correct...if your IP is 192.168.1.231 the pool must be 2 - 230 and the next pool starts 232 -255...

 

[yonoss]

I've assigned a new IP to the proxmox server, outside of the ips pool DHCP is using. Thanks !

 

[yonoss]

Shouldn't proxmox installation UI be enhanced to make it clear that it is allocating a static IP and not a dynamic IP, as it currently gives the impression by picking the ip automatically ?

 

[Nollimox]

Shouldn't proxmox installation UI be enhanced to make it clear that it is allocating a static IP and not a dynamic IP, as it currently gives the impression by picking the ip automatically ?

It does state that as it's a server and all servers require static IP...

 

[yonoss]

I've assigned a static IP outside of my DHCP server ips range but I still see this behavior. Especially after a cold start (server was down for a couple of hours)


The new IP is 192.168.1.6 , while the DHCP server IPs range is from .10 to .254

Note that I also have some other servers (not proxmox) which are making use of static IPs in the same range .2 - .10, and I don't see this problem with them.

 

[Nollimox]

I've assigned a static IP outside of my DHCP server ips range but I still see this behavior. Especially after a cold start (server was down for a couple of hours)


The new IP is 192.168.1.6 , while the DHCP server IPs range is from .10 to .254

Note that I also have some other servers (not proxmox) which are making use of static IPs in the same range .2 - .10, and I don't see this problem with them.View attachment 61705View attachment 61706



View attachment 61704

Notice that you only have one NIC...a cloud setup?

 

[cwt]

Is PVE connected to a managed switch? If so, can you check the switch‘ logs for any errors?

 

[ubu]

With not visible on the network I assume you mea you cannot ping the address of your proxmox Host, correct?
So while you ping the proxmox Host IP from your client check with tcpdump -i eno1 icmp and tcpdump -i vmbr0 icmp on your proxmox if it is receiving the ping packages.

Try to ping your roter from the proxmox, does that work, if yes can you ping the proxmox server after that?

Send output of ip addr list and ip route list

Send output of ethtool eno1

 

[yonoss]

Notice that you only have one NIC...a cloud setup?

No. It is a standalone server on my home network. Although the server has 2 physical NICs, I've only connected 1.

 

[yonoss]

Is PVE connected to a managed switch? If so, can you check the switch‘ logs for any errors?

It is connected to a home wifi router. I've checked the logs but I'm not seeing anything clearly referencing the server ip or mac address

 

[yonoss]

With not visible on the network I assume you mea you cannot ping the address of your proxmox Host, correct?
So while you ping the proxmox Host IP from your client check with tcpdump -i eno1 icmp and tcpdump -i vmbr0 icmp on your proxmox if it is receiving the ping packages.

Try to ping your roter from the proxmox, does that work, if yes can you ping the proxmox server after that?

Send output of ip addr list and ip route list

Send output of ethtool eno1

Regarding your first question the answer is : yes that is correct. The Proxmox web console is also not reachable.

Pinging the router from my proxmox server when the problem occurs, doesn't work either.

Tcpdump receives nothing when I'm pinging the proxmox server:

Pinging 192.168.1.6 with 32 bytes of data:
Reply from 192.168.1.216: Destination host unreachable.
Reply from 192.168.1.216: Destination host unreachable.
Reply from 192.168.1.216: Destination host unreachable.

I've attached the screenshots for the other stuff you've asked for

 

[ubu]

ok, so your nic is not started properly.

Please post output of
1. systemctl status networking
(while it is broken)

then do a systemctl restart networking and systemctl status networking again

 

[ubu]

and a journalctl -b -p 3

 

[bbgeek17]

At least two of your screenshots show "NO CARRIER" and "NO LINK" detected on your eno1 interface. That's a physical layer issue, unlikely to have anything to do with DHCP distribution.

Try a different cable, different port and finally different NIC. Even thousand dollar Mellanox NIC sometimes are DoA. It happens. Or the chipset on the NIC is no (longer) compatible with the Kernel, or support got broken.

The fastest way to resolution is by swapping the parts.

You can also try to throw LINKDELAY in your config file, although I am not sure what the equivalent of it is netplan format.
https://support.hpe.com/hpesc/public/docDisplay?docId=sf000059866en_us&docLocale=en_US

Good luck

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox