[SOLVED] Cannot ping server in the same network

lpwevers · 03-13-2017, 03:22 AM

Dear experts,

I've come across a very interesting challenge. Hopefully has any of you seen this issue before and knows the solution. In a rack I have a server A with IP 172.29.40.5/24 which connects to the great outdoors via 172.29.40.1. That all works fine.

Now, I've added 2 new servers (B & C) to the rack, using 172.29.40.20/24 and 172.29.40.22/24. They all connect to the same physical unmanaged switch. Now this is where it get's interesting. From B I can ping C and vice versa, but not A, nor the gateway. I've checked and double checked that all network settings are as they should. And since B & C can ping each other, I'm convinced that there's nothing wrong with the cables or the switch. But just to eliminate that possibility I've also tried different ports on the switch and even a complete new switch. Also tried different NICs in the machines, all to no avail. I'm not using VLANs, so that cannot be the issue either.

I've also checked that I don't have an issue like duplicate mac addresses on any of the nics. If I run 'arp -a' on B and C I see that they have addresses for each other, but for A and the gateway they say that the response is incomplete. So I added the address of A manually to the arp table of B & C but that didn't work either.

Just to rule out any issues with the OS (B & C were pre-installed at another location using a different network configuration and they worked just fine there) I've booted using a live CD just to get a clean OS and try from there, but again, same result.

So now I'm at a complete loss as what I can try next. Any suggestions are greatly appreciated.

pan64 · 03-13-2017, 03:49 AM

have you tried traceroute?
firewall settings?

lpwevers · 03-13-2017, 03:58 AM

Quote:

Originally Posted by pan64

have you tried traceroute?
firewall settings?

Traceroute unfortunately has the same results as ping. Host cannot be reached. There are no firewalls in place on the machines. All are on the internal network.

ferrari · 03-13-2017, 04:11 AM

For each host, if you flush the arp cache

Code:

ip -s -s neigh flush all

and then do

Code:

nmap -sP 172.29.40.0/24

what is then reported in the arp cache of the respective host?

Code:

arp -a

lpwevers · 03-13-2017, 05:53 AM

Hi,

Thanks for 5he reply. Unfortunately I have no nmap on the server. And as long as it has no connection to the internet, I'm afraid that I cannot access the repositories for installation.

I'll see if I can get it to work with a live cd.

lpwevers · 03-13-2017, 06:38 AM

Ok, the live-CD comes with nmap. I get the following results:

Code:

# ip -s -s neigh flush all
Nothing to flush.

# nmap -sP 172.29.40.0/24
Starting nmap 3.81 ( http://insecure.org/nmap ) at 2017-03-13 11:30 CET
Host 172.29.40.20 appears to be up.
MAC Address: 10:60:4B:A0:57:9C (Unknown)
Host 172.29.40.22 appears to be up.
Host 172.29.40.26 appears to be up.
MAC Address: 3C:4A:92:76:F5:5E (Unknown)
Nmap finished: 256 IP addresses (3 hosts up) scanned in 38.503 seconds

# arp -a
 (172.29.40.35) at <incomplete> on eth0
 (172.29.40.199) at <incomplete> on eth0
 (172.29.40.213) at <incomplete> on eth0
 (172.29.40.127) at <incomplete> on eth0
 (172.29.40.113) at <incomplete> on eth0
 (172.29.40.252) at <incomplete> on eth0
 (172.29.40.128) at <incomplete> on eth0
 (172.29.40.89) at <incomplete> on eth0
 (172.29.40.130) at <incomplete> on eth0
 (172.29.40.180) at <incomplete> on eth0
 (172.29.40.218) at <incomplete> on eth0
 (172.29.40.87) at <incomplete> on eth0
 (172.29.40.105) at <incomplete> on eth0

The reachable other address 172.29.40.26 is the iLo of the server.

pan64 · 03-13-2017, 08:02 AM

I would run it on A, B and C too. I still think it is a configuration problem. Also you can try wireshark to check low level traffic.

lpwevers · 03-13-2017, 11:13 AM

Ok, finally got this working. These are servers that are hosted in a co-location datacenter. Turned out that the hosting provider actually was using VLANs in a sneaky way. Never mind I added the VLAN tags and now it's working.

Thanks for all the help.

ferrari · 03-13-2017, 02:49 PM

Information like this is always vital..

...anyway, good to know that it is working as expected now.