I have a test lab set up running a virtualized windows 2003 domain on a CentOS 5.3(xen) host and am experiencing connectivity problems with guests running on other hosts that are part of the same domain.

Here's the setup:

On Computer A I have CentOS 5.3 running as the host and have virtualized windows 2003 servers for a primary domain controller, a backup domain controller and an exchange server. The primary domain controller also acts as a WINS and dns server. The windows domain appears on a separate subnet from my company's corporate network. Connectivity to any of the virtualized guests on Computer A is fine (remote desktop, ping, what have you).

I have another host computer (Computer B) that also has a virtualized Windows 2003 server guest that is part of the same domain. However, connectivity to that guest is flaky at best. I continuously get at least 60% packet loss when I try to ping the guest, and due to that flakiness I can not access any of the services that it runs (remote desktop, web).

Now here's the interesting part. It seems to affect only machines running on a different computer than the domain controller that are in the same domain. On Computer B there is another Windows 2003 guest that is not part of the test domain and is on my corporate network. There's no connectivity issues with that guest machine.

The problem does not seem to be specific to Computer B either. I created a test VM on my local computer within the test domain and it exhibits the same behaviour as the guest in Computer B.

A couple of items to note: - Host OS on both Computer A and B are the same CentOS 5.3 64 bit - Guest OS is Windows 2003 64 bit and 32 bit (the guest on Computer B is 32 bit) - Guest OSes are all up to date (as of Monday) - Host OS on Computer A was upgraded from CentOS 5.2 to 5.3

Update: Sorry I did not follow up with the comments from below. Computer A and B have been moved to their own dedicated switch and the problem has gone away. I'm not sure what the underlying problem(s) were though

  • Are there incrementing collisions or errors on your ethernet interfaces? – James Cape Jun 25 '09 at 2:18
  • When you ping a computer that you are seeing packet loss with, does it look normal? I've seen clock issues with VMs that cause bad networking behavior. – dmo Jun 26 '09 at 19:11
  • Just to be clear - Does the non-domain guest machine on host B have any problems communicating with the guest DC on host A? – John Gardeniers Jul 3 '09 at 22:54

Have install xenpv driver from redhat

BUGZILLA : https://bugzilla.redhat.com/show_bug.cgi?id=469767

Errata : http://rhn.redhat.com/errata/RHEA-2009-1583.html ( Package: xenpv-win-1.1.0-2.el5.noarch.rpm) .

its need restart to your server win2k3 server it will install para virtual driver for your win 2k3 server.

| improve this answer | |

Based on your success when you moved to a single switch, I'd suggest that you have not yet fixed the problem, just hidden the symptoms.

Whilst on your private switch, I suggest running wire shark and looking carefully at the network traffic, and then attempt to explain every single packet you see on the wire.

The end result will be a much better understanding of Microsoft's use of ports 135, 137-139 and port 445 which are used in file sharing, and you are likely to see packets that explain why you had problems before hand.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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