Weird DHCP client issue

[dgingeri]

OK, to start off, here's the environment:

Windows Server 2008 R2 DNS/DHCP server with DDNS updating from the DHCP server service
ESXi Host: ESXi 5.0 on a Dell PE1950
VMs: 2X CentOS 6.2, one network connection on each

Here's the problem:
For some odd reason, when these VMs come up, they request a total of 3 DHCP leases, one with the MAC address and no name, so DDNS is not updated, one with a weird 8 hex digit unique identifier that is definitely not the MAC address, and one more with that same identifier and some extra digits that seem random. The second and third requests both include the system name so DDNS updates with those, at the same time. The machine actually uses the first one. The DNS server then presents the wrong IP address when trying to reach it from outside the lab.

I don't know if it is a difference between Linux and Windows interpreting the DHCP standards, but this doesn't happen with any of our CentOS5/ RHEL5 VMs, or even our Ubuntu or SLES VMs. Those work great and update DDNS just fine. OF course, all the Windows VMs work fine.

Does anyone have a clue what's going on?

 

[dgingeri]

Does anyone actually read this forum?

 

[Ijack]

You are unlikely to get help if you show an impatient attitude. You might be better posting in a specialized Centos forum with such a paticularized problem. My guess is that it's to do with the configuration of your VM, but as I can't see that configuration it's just a guess.

Be patient and try not to be rude to potential readers; you will get better results that way. I'd suggest that you wait at least a day, at the very least, before you consider bumping a post; many would think that generous and that it is always presumtuous to bump your posts.

 

[dgingeri]

Sorry, I had the user ask me about this issue 6 times today after I posted this, and the CentOS 6 networking forum posting I put up got a similar lack of response. I've been searching for a fix, or at least a workaround for this for two weeks. I'm impatient because I'm being pushed for answers.

I've tried contacting Red Hat support on this, as it affects our Red Hat 6 VMs as well, and basically just got blown off by them as "something's wrong with your DHCP server." Of course, developers never admit when there's a bug in their software. I know that all too well from working in a software test lab.

This is a last ditch effort. I've posted it in the CentOS forums, then this forum an hour later, then HardOCP's forum an hour after that. I'm totally out of ideas.

 

[skittle]

if you want immidiate answeres look to IRC. ask on #centos on freenode

 

[Ijack]

If I were you I would try installing the same software on a physical machine to see if you get the same problem. That way at least you will know who to shout at. Without the ability to do any tests I would still look to the VM configuration as being a likely cause which I would want to eliminate with such a test.

 

[Ijack]

Just a quick addition to my previous post. If this is causing real issues, why not just assign fixed IP addresses to these servers?

 

[dgingeri]

We ended up doing that, late yesterday, but it's interfering with our tests. Supposedly, our customer is using DHCP and Dynamic DNS for their CentOS6 machines and running our software. DHCP and Dynamic DNS are working for them (so they say) but the software is having a different issue.

I'm now betting this is the problem they're actually running into. Assigning static IP will probably fix their problems too, since our software is heavily dependent on DNS. It's probably this same Dynamic DNS issue that only affects CentOS/RHEL 6.

 

[dgingeri]

Best answer selected by dgingeri.