Networking help needed with 2 Unix servers

[Thilton]

Hello all,

I'm a M$ guy having to connect two Sun servers to our network and I'm running into a bit of a strange problem. I've got both servers (A and B) connected to the network and they can ping any other server on the network, except each other. Server A cannot ping Server B and Server B cannot ping server A.

Now here's the strange part. If I go to any other workstation in my network I can ping Server A but not Server B. However, if I reboot Server B then I can ping server B and not Server A. In fact, when I check the arp table in my firewall only one server will register at a time.

Has ANYONE seen ANYTHING like this before? I'm scratching my head over this and just don't know what else to do here. Any thoughts?

Thanks!

Troy

 

[julianbarnett]

Check 'ifconfig -a' on both servers - both the IP and Ether (MAC) addresses should be different. Sounds like they might have the same IP address.

Whats the network topology? Are they on the same subnet and non-routing - i.e. not sending ARP redirects to each other.

 

[Thilton]

I know for a fact that they have different mac and IP addresses. I have my Unix guy here today so we're going to go through and reset the IP configurations just to be sure.

They are on the same subnet though.

Troy

 

[Thilton]

OK, We've reset all the network configurations and we're still getting the same thing. Only one server will function properly on the network. They are both members of the domain(we're running Active Directory) and have unique IPs and MAC addresess, they're both on the same subnet and share the same gateway.

Can anyone provide any other information on this? We're desperate.

 

[rblue]

Check the /etc/defaultdomain file and make sure that the entry in it matches the domain that the host is in. It should be fully qualified, without the hostname, e.g., domain.comany.net

Next check /etc/defaultrouter and make sure that the entries there point to the correct IP address(es) of your DNS server(s).

Make sure the subnet is correct in /etc/netmasks

Finally, make sure that the hostname in the /etc/nodename and /etc/hostname.le0 file is the short hostname, not the FQDN.

Also helps to see if the other NT servers are in the /etc/hosts file.

Do this on both hosts.

Also, I KNOW you're not running DHCP on these, but just in case - get them off DHCP. Sun just doesn't do it well.

 

[Annihilannic]

Do the Sun systems have multiple NICs? If so, touch /etc/notrouter and kill the routing daemon routed to eliminate that possibility... it's probably worth doing this regardless.

Annihilannic.

 

[Thilton]

Thanks folks. Hey rblue, You are correct. These are NOT using DHCP.

I'll have my Sun guy take a look at these.

Thank you all very much for your input!!!

Troy

 

[Thilton]

rblue,

I checked the defaultdomain file and found that it didn't exist on either server. I next checked the default router file and fond that one server had the file but it pointed to the gateway of our firewall. There were no instances of DNS server IPs in that file.

Next I checked the netmask and it was correct on both servers as was the nodename. There was no hostname.le0 file on either server. The hosts file has 2 IPs: 127.0.0.1 and the IP address of the server itself. That's it.

You mentioned that the defaultrouter file should point to my DNS servers. I'm not certain if this is part of the problem as I'm not trying top use the FQDN, just the IPs. In my thinking, since I'm only trying to ping the serves the DNS servers don't really come into play at this point. But what about the hostname.le0 file? What's that for? Please forgive the ignorance of my questions as I know nothing about Unix.

Troy

 

[julianbarnett]

/etc/defaultdomain is a red herring - this is used when setting up NIS.
hostname.le0 is used by the server to configure interfaces on reboot - the file is named according to your interface name i.e. hostname.hme0 or whatever. Simply contains a name which netconfig matches in /etc/hosts to set the IP address. The netmask on reboot is defaulted to whatever class the address is in unless it is overidden by an entry in /etc/netmasks. The default gateway is set from from /etc/defaultrouter.

All you really need to set up basic networking is

/etc/hosts
#
# Internet host table
#
127.0.0.1 localhost
172.16.64.2 myhost myhost.mydomain.com loghost

/etc/hostname.hme0 <-- Your interface name here
myhost

/etc/netmasks
255.255.255.0 <-- If subnetting an X-class network

/etc/defaultrouter
172.16.64.1 <-- Your default gateway

Can you post output of 'ifconfig -a', 'netstat -rn' and 'arp -a' from both servers.

regards Julian

 

[Thilton]

Problem resolved! The cause was the monitor and keyboard. I only have one monitor and keyboard to handle both servers. If I booted the servers with the monitor and keyboard connected, then disconnected it the server would drop from the network until they were reconnected and rebooted. To resolve the issue I simply rebooted both servers without the monitor and keyboard. problem solved.

Thanks for all the help and suggestions!

Troy