Remote Session Lost During Validation
Hi,
I have two clusters built using Windows 2008 - two nodes each. The first is VM and for test only. I run the cluster validation and have no issues other then a fibre error. (The fibre issue is not the problem.) I'm able to run the test from either node without being disconnected.
However, when I run the cluster validation on either node of the physical cluster, my remote session is disconnected until the validation is complete. I am able to reconnect to the session and see the report. During the test, a ping of the node where the test is run fails.
The virtual cluster uses fixed IPs while the physical cluster uses DCHP. Is this the cause of the disconnect? Is there something else that is I should be checking?
Thanks,
Randy
답변
- These servers are getting their IPs via DHCP? I don't know of any documentation that says this is not supported, but that is a really bad idea even if it is supported. I'm fairly certain that this is your problem. Can you configure them with static IPs and run your validation again?
- 답변으로 표시됨Randy in Marin 2009년 11월 18일 수요일 오후 9:27
모든 응답
- I'm not sure why this is happening, but perhaps doing it from the console will give you more information.
- I have more information.
1) My session is disconnected only if I elect to take the MSDTC application offline during one of the storage tests. (On the VM version, I take the MSDTC and SQL Server applications offline without a disconnect.) There are no disconnects or validation errors if MSDTC remains online.
2) I noticed on node B that the Component Service would lock up when I tried to view the settings. I added msdtc.exe as an exclusion in the firewall to fix that. Because the firewall is disabled on the VM version, I went ahead and disabled the firewall completely on the physical nodes just to see if it might be related to the disconnect problem. I rebooted both nodes - one at a time - and re-tested. The issue remains.
3) There are two disabled network adapters on each physical node. (There are two enabled for the public and private networks.)
4) I noticed that at times that it appeared that I could remote into only one node at a time. Actually, for some reason, when I remoted to node A, I would be connected to node B. This caused the disconnect with my existing session to node B. Both names pinged with the correct names, so why did it get redirected?
5) I disconnected from all sessions and remoted into the node (node A) that was being "redirected". I was able to remote into it and got the initial configuration screen - odd. I tried running the validation report from this node and elected to take MSDTC offline. As soon as the storgae was taken offline, the remote session to the other node (node B) was non-responsive. After the test completed, the session to node B was re-established. In addition, I get validation errors. For example.
List BIOS information from each node.An error occurred while executing the test. There was an error getting information about the BIOS of the nodes. Unable to connect to SQLEJUS2B.co.marin.ca.us via WMI. This may be due to networking issues or firewall configuration on SQLEJUS2B.co.marin.ca.us. The RPC server is unavailable. (Exception from HRESULT: 0x800706BA)
Validate that all the nodes have the same domain, domain role, and organizational unit.An error occurred while executing the test. Failed to initialize the Configuration tests. Unable to connect to SQLEJUS2B.co.marin.ca.us via WMI. This may be due to networking issues or firewall configuration on SQLEJUS2B.co.marin.ca.us. The RPC server is unavailable. (Exception from HRESULT: 0x800706BA)
I don't know what to do at this point. Could this be DCHP related? Should I continue to worry that I have a configuration issue?- 편집됨Randy in Marin 2009년 11월 17일 화요일 오후 9:48typo
- These servers are getting their IPs via DHCP? I don't know of any documentation that says this is not supported, but that is a really bad idea even if it is supported. I'm fairly certain that this is your problem. Can you configure them with static IPs and run your validation again?
- 답변으로 표시됨Randy in Marin 2009년 11월 18일 수요일 오후 9:27
Hi Tome. Our server team will be checking DNS. I have also requested static IPs. Once this is done, I'll give it another try and report the results. Thanks.
- Well, the problem was resolved. We still use DHCP for the node and cluster IPs, but we now use a static IP for MSDTC. Our DNS entries needed cleaning and syncing. One of our DNS servers had the wrong IPs. I think that's why I would get redirected to the wrong machine at times. I don't know if it was the static IP for MSDTC, the DNS cleanup, or both that finally fixed the issue.
PS Our server team did not want to use static IPs. They did set the lease to not expire on those used.
Thanks,
Randy
