[SOLVED] New Installation - no network

[jonathanku]

Hi folks, can you help me troubleshoot the network for a new installation of Proxmox VE 8.2.

I can't access the browser page from another PC after installation. Can't ping to or from the proxmox host. Interface is now "UP" I think after I put a network switch between proxmox and the main router (previously seemed to need that setup -- hardware issue maybe?)

Here's my interfaces and ip addr output




Interestingly, when I did this nmap scan from my other PC, it seems to have found the FQDN (ending .co.uk) I gave it during installation

 

[gfngfn256]

1. You used a CIDR of /24 when you setup Proxmox, but on your nmap scan you used a /32 CIDR
2. On your nmap scan you added a CIDR to a specific (single) IP address - that is not the intention of the CIDR addition to nmap. See here.

Assuming the CIDR is really /24 - use one of the following to scan for your PVE instance:

nmap 192.176.1.0 /24 or look specifically for nmap 192.176.1.64

Also what does the "another PC"'s local IP show: ip a

 

[jonathanku]

Firstly - ignore my point about it finding the FQDN - that came from my other PC's hosts file which I obviously had created a few years back - and have just reused the same IP this time round.

From your nmap link:

"The largest value for IPv4 is /32, which scans just the named host or IPaddress because all address bits are fixed."

ip a from the other PC:

1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host
       valid_lft forever preferred_lft forever
2: eno1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP group default qlen 1000
    link/ether 70:85:c2:8f:08:f4 brd ff:ff:ff:ff:ff:ff
    inet 192.168.1.46/24 brd 192.168.1.255 scope global dynamic noprefixroute eno1
       valid_lft 86271sec preferred_lft 86271sec
    inet6 2a00:23c5:2782:4e01:c072:d7b5:1659:4ac/64 scope global temporary dynamic
       valid_lft 604674sec preferred_lft 86098sec
    inet6 2a00:23c5:2782:4e01:779e:921e:a9b9:e279/64 scope global dynamic mngtmpaddr noprefixroute
       valid_lft 315359988sec preferred_lft 315359988sec
    inet6 fdaa:bbcc:ddee:0:c072:d7b5:1659:4ac/64 scope global temporary dynamic
       valid_lft 604674sec preferred_lft 86098sec
    inet6 fdaa:bbcc:ddee:0:ce3f:2684:85e8:dc78/64 scope global mngtmpaddr noprefixroute
       valid_lft forever preferred_lft forever
    inet6 fe80::6475:685a:ca07:4463/64 scope link noprefixroute
       valid_lft forever preferred_lft forever

 

[gfngfn256]

I put a network switch between proxmox and the main router

Plug in the other PC in that switch too. (You've probably got 2 different NWs created; one by the router & one by the switch - all sorts of mess). Then try ping 192.168.1.64 from that other PC connected to the same switch.

IDK your router (& PVE setup) but you need to address why it won't work when you attach the PVE directly to the router. Some address reservation/mac limiting problem?

 

[gfngfn256]

One other thing I notice on your original post image, is that in the PVE interfaces setup you've got the gateway as 192.168.1.254 - while it is possible that it is correct - the most default one for that NW is 192.168.1.1 - just checking.
You could check the GW from the other PC with ip route | grep default

 

[jonathanku]

Okay - plugging PC & Proxmox mini-server into the same switch, and I can now ping proxmox from the PC, and access the browser page!

I now wonder if my entire network - as you say - is in all sorts of mess.

Warning - the scribble below might make your eyes bleed if you're a professional... (EDIT: updated diagram - slightly more tolerable) - that's how it looks after your suggestion (PC and Proxmox server into same switch).

Gateway is correct as 192.168.1.254 - that's how the "BT Smart Hub" internet modem/router was configured out the box.

Interestingly now that it has come to life, I can also reach the Proxmox server from my phone via wireless to the BT Router - so it's not just PC-to-Proxmox connectivity via that 4-port switch alone. I'm confused.

 

[jonathanku]

I've tried a couple of things - not sure if it's a valid test or not... but I can plug the Proxmox server into any of the 4 port switch ports and it comes online. But if I plug into the second wall port directly (so it's going to a different port on the 8-port switch) it doesn't seem to get on the network. However, when connected to the 4-port switch, I can move that to the second wall port (i.e. that different port on the 8 port switch) and it does get on the network fine. I've also tried plugging the proxmox server directly into the BT Hub (i.e. the main router) and it doesn't get on the network.

 

[gfngfn256]

There are 4 general possibilities for your NW "problems":

1. Your PVE server is setup/configured that it only connects with that 4 port switch.
2. Your router/switches are setup/configured that only the 4 port switch allows/manages to correctly connect to the PVE server.
3. Your physical NIC on the PVE server is somehow only compatible with the 4 port switch.
4. Your PVE server is in fact connecting to the other ports/switches/router - but its somehow isolated when you test ping it based on the setup.

Be aware that you could have a number of the above.

I don't know your NW setup of the above & I'm sorry I really don't have much time to help you - but you'll need to go through all of the above systematically.

Just my hunch - you have probably got conflicting NWs/GWs/addresses setup with those router/switches. Also your router is also a Wi-Fi AP (going by the image above + your phone mention) that may also bring with it more of the above + complexity (is that PC Wi-Fi enabled, is your PVE server also Wi-Fi enabled?).

Anyway good luck to you.

 

[jonathanku]

Thank you

 

[sw-omit]

One other thing to add that it could be (although I kinda of doubt it in a at-home-setup), would be bad quality network-cables or cables that are too long.

Between each "active" device, you can normally have about 100 meter / 300 feet of network-cable, but if the cable is damaged or of a bad quality, it could be less before you get degredation at least and no connection at worst.

In your test with the 4 ports switch and second wall-outlet, were you using the cable that used to be between switch and server, or between switch and wall to connect wall and server?
if you want to check if it might be a cable problem, probably the easiest way would be to try to use the cable that you used to connect switch to wall to now connect server to wall. If you were already doing that, then this is probably not the issue.

 

[jonathanku]

Thanks for the input. Yes, I ruled out most of my cable issues early on. One cable runs under the floor to a wall port, and I wondered if mice had nibbled it - but I then tested with a working short cable straight into the router, and still the same. It's not an intermittent fault either - it seems to decisively not work or work fine.

Good news though - when i started hunting for issues with my hardware (Shuttle XS35 v2, quite old) I found someone with similar issues. And changing this setting worked for him - and it immediately caused my network connection to come back up (though not "eth0" in my case):

ethtool -s eth0 speed 100 duplex full autoneg off

So there's something about auto-negotiation, I think, not working right.

Just need to automatically run that on boot up, somehow.

EDIT: solution I've settled on (for now) is a new script under /etc/network/interfaces.d/ to run the ethtool command:

auto lo
iface enp2s0f5 inet manual

pre-up /sbin/ethtool -s enp2s0f5 speed 100 duplex full autoneg off
up ip link set dev enp2s0f5 up
down ip link set dev enp2s0f5 down

 

[sw-omit]

Does does mean you're only running at a 100 Mb/s speed though of course, instead of a 1 Gb/s, but glad to hear you came to a workable and stable solution (at least till you can upgrade/replace the network-card ).

Question though, if you do the same command with a speed of 1000 (still full duplex and autoneg off) does that work?

Also, thinking about it some more, is that 4 port switch you were using perhaps an (old) 100 Mb/s switch and the 8 port switch a 1Gb/s switch?


Edit: Quick look at the specs-page, looks like the card does/should support a gigabit connection:
https://global.shuttle.com/products/productsSpec?pn=XS35GT V2

 

[jonathanku]

Also, thinking about it some more, is that 4 port switch you were using perhaps an (old) 100 Mb/s switch and the 8 port switch a 1Gb/s switch?

Interesting - just checked, and yes, it's a 10/100Mbps switch!

To be honest, once it started working, I was happy to just go with 100 and get on with my project - I probably won't need more bandwidth. But it's a good question - was it setting it to 100 which solved it, or disabling the auto-negotiation. I'll try running the same command but 1000 instead of 100 and see what happens.