Take the 2-minute tour ×
Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

On a small college campus we have a VLAN setup for the computer labs. These machines get assigned IP addresses in the 192.168.7.xxx range. In the server room, all of the server are on the default VLAN and assigned an IP address in the 10.1.1.xxx range.

For the most part this works, but the lab machines are unable to connect to one of the servers. They can't even ping it. They can talk to other servers on the same switch as this server just fine. At first I thought it might be a vlan issue, but I changed the server port vlan to match other known-working ports with no effect.

Any ideas?


I don't think it's a firewall issue — non-lab machines work just fine.


The server also cannot ping the labs, so it's definitely a problem in both directions.


I just tried tagging a couple lab ports for the server's vlan (lab vlan is untagged), but that didn't work (didn't think it would, but I had to try).


To be clear: the lab can ping another server on the same switch and vlan as the problem server. The other server it not on the same switch or vlan as the lab machines. So I know I can send pings through the network, including all trunks, as far as the destination switch. It's got to be something different with the server or port. Except it's not just this one server.


The problem gets weirder. I have a third server on the same switch as the other two. This server has multiple NICs. I can talk to the server on one nic but not the other. It is also a VM host, and I can't talk to any of the hosted VMs.

share|improve this question

5 Answers 5

You probably have a router between the VLANs, right?

I'd start from the router and establish that you can ping from there both ways. If the router can ping the server and the lab machine, there's (possibly) a routing issue on the server. You say you've checked the default gateway, but is there a more specific route?

If you can ping the lab machine, but not the server from the router, it could be netmask (maybe), it could be a faulty cable, so probably worth swapping the cable out for a known-good cable. It might even be a MAC address collition (unlikely, but I have seen it, use the CAM in the server switch to see if you're seeing any MAC on more than one port).

share|improve this answer

Perhaps a firewall rules on the router or server itself needs to be added? If you can ping other servers, I would guess routing is fine if they are all in the same network.

share|improve this answer

Maybe the routing table on this machine is messed up? Or maybe it has a custon firewall which drops incoming packets?

If you have access on the server, check the firewall and also test the connection the other way around (from the server to the client) and see if there's connectivity.

share|improve this answer
    
I don't think it's a firewall problem - it talks to other machines just fine. It is probably routing though. –  Joel Coel Mar 12 '10 at 16:00

Routing and firewall seems a little too high level for this issue. You say servers connected on the same switch will ping, but ones connected to other switches do not. I would say you need to trunk your VLANs on the switches.

Check all your trunks and make sure that the necessary VLANs are added. If you run VTP, make sure you added the new lab VLAN to VTP using the VTP server.

The reason I do not think its a routing issue is because if two devices exist in different networks they always have to talk to a gateway before they can talk to each other. Doesn't matter if they are connected to the same switch or not.

So in the case of the server and lab computer, they are in different VLANs. So if one tries to ping the other, that ping has to go up to a layer 3 address first (the gateway) before the ping will respond.

The OP states that a ping from the lab computer to the server on the same switch works fine, so routing should not be an issue.

share|improve this answer
1  
To be clear: the lab can ping another server on the same switch and vlan as the problem server, not on the same switch or vlan as the lab machine. So I know I can send pings through the network, including all trunks, as far as the destination switch. It's got to be something different with the server. –  Joel Coel Mar 12 '10 at 18:14
    
That would make sense as its the only other possibility... –  Webs Mar 12 '10 at 21:19

Well, check the default gateway of the problem server, compare it with that of other servers in the same vlans. Your ping won't work, if the default gateway of the problem server is different.

In such a case, the problem server can talk to other machines in the same VLAN, but can't talk to machines in other VLANs.

share|improve this answer
    
The only difference in gateways was the metric (the working server's metric was '1', but the bad server's metric was 'auto'). Changing the bad to match the good didn't help. –  Joel Coel Mar 23 '10 at 20:21

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.