Two servers at different locations share same private IP

[md03]

Greetings,

I have two servers at different locations, both on private subnets. Both servers share that IP address 192.168.0.6. The first server auto-discovers fine and all is well. After installing the 3.2.5 agent on the second server it does an autodiscovery but then attempts to update the console database with the new information matched to the previously configured server's name and IP, I assume this is because of the shared IP address?

I tried finding something in the agent.properties to override this behavior with no luck. Any input on how to fix? I will definitely run into this scenario again.

Regards,
Mike

[excowboy]

Mike,

could try to setup a different port for the shared IP-Adress, for example default port 2144 for the first server and 22144 for the second one ?
Run hq-agent.sh setup on one of the servers to reconfigure the port.
Do the servers also share the FQDN ?

hope this helps,
Mirko

[md03]

Sorry for the delay. I did get to try using a different port but when doing so the agent and server are unable to communicate during the agent setup process.

In answer to your question of FQDN, that is a unique charecteristic.

I've searched the forums a bit but don't see anything pertaining to the agent property that the database is using as a primary index key. Would be nice to be able to make that be the MAC address instead of an interface primary IP.

Note: Agents are running Windows Server 2003 if anyone has any other ideas to try. I may just upgrade to 4.0 and see what the result is. If so I will repost with info.

Regards,
Mike

[vreikine]

This sounds like NAT problem. There is no way you can have 2 differrent systems with the same IP address on the regular IP network without some sort of Network Address Translation. I am not sure about suggestion to run agent on different ports - you have problems on IP layer , and TCP is one level up. Try traceroute from server to that IP address , I don't think it would work.

[edan_hyperic]

> This sounds like NAT problem. There is no way you can
> have 2 differrent systems with the same IP address on
> the regular IP network without some sort of Network
> Address Translation. I am not sure about suggestion
> to run agent on different ports - you have problems
> on IP layer , and TCP is one level up. Try traceroute
> from server to that IP address , I don't think it
> would work.

+1

[excowboy]

>I am not sure about suggestion to run agent on different ports - you have problems on IP layer , and TCP > is one level up.

The usage of two different ports is relevant for Hyperic HQ, because the port is part of the unique identifier of an Agent on a platform. If I remember correctly there is a forum post from a Hyperic Engineer saying that it's part of the identifier. Sorry, but I can not find it right now.

Cheers,
Mirko

[ALWalsh]

Hey,

I'm experiencing the same problem. One thing that I have noticed is that the host name and the ip address (private) of the servers that are been mismatched are identical. Perhaps it is the combination of these two variables that makes up the resource name within hyperic???

A solution might be to change the DHCP scope of one of the networks that are been mismatched???

Don't really want to do this, so if anyone has any other ideas I would very much appreciate it?

Thanks

Alan