How to access Hyper-V VMs on a Cluster Shared volume when Cluster is offline
Crisis:
I have a 2 node Win 2008 R2 Failover Cluster which I'm having difficulties getting online.
It's related to AD difficulties as far as I can tell.
For some reason I cannot Access the Cluster Shared Volume even though its listed as online in Failover Cluster Manager.
I really need to get one of the VMs stored on the CSV, and I'm wondering how this can be done?
For example: is it possible to take the CSV offline, remove it as a cluster shared volume, so that the "reserved" status the disk has in Disk Management will be gone? And then it can be accessed normally?
Or: what happens if I destroy the cluster (I'm in the process of moving the VMs to a New cluster in a New AD anyway).
Will this remove the reservation on the volume which now is the CSV, and let me Access the VMs on this volume from Hyper-V Manager? So I can continue to run them for a short while as I'm moving them one by one to their New home?
Any suggestions are welcome!
I have a 2 node Win 2008 R2 Failover Cluster which I'm having difficulties getting online.
It's related to AD difficulties as far as I can tell.
For some reason I cannot Access the Cluster Shared Volume even though its listed as online in Failover Cluster Manager.
I really need to get one of the VMs stored on the CSV, and I'm wondering how this can be done?
For example: is it possible to take the CSV offline, remove it as a cluster shared volume, so that the "reserved" status the disk has in Disk Management will be gone? And then it can be accessed normally?
Or: what happens if I destroy the cluster (I'm in the process of moving the VMs to a New cluster in a New AD anyway).
Will this remove the reservation on the volume which now is the CSV, and let me Access the VMs on this volume from Hyper-V Manager? So I can continue to run them for a short while as I'm moving them one by one to their New home?
Any suggestions are welcome!
ASKER
Hi, and thanks for helping!
- The CSV is on a SAN Device
- Fibre
- Yes, the node is online
- No, do not have a DC not on the cluster (anymore, used to have)
- No, haven't tried the validation wizard
- The CSV is on a SAN Device
- Fibre
- Yes, the node is online
- No, do not have a DC not on the cluster (anymore, used to have)
- No, haven't tried the validation wizard
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Hi, sorry that I haven't been able to give an update sooner.
I had to work continously through the night and on Friday before I managed to solve the issue(s).
To summarize:
I managed to remove the CSV from the cluster, and Access the volume from Windows Explorer on one of the Hyper-V hosts.
This allowed med to copy the VHD-file for the VM I needed (the DC) to the local disk and create a new VM based on this file.
After bringing the DC online, when I tried to log on to the Hyper-V hosts, I got an error with the Security/trust relationship between this computer and the domain.
Then I had to log on with local admin, remove the server from the AD domain, reboot, re-join the domain and reboot.
Not a big problem, except that one of the 2 servers didn't have the standard password for the local admin account.
This was around 4 AM, didn't make me very happy :)
When I finally got around this problem and had gotten both Hyper-V hosts re-joined to the domain, with the DC running, I could get on with getting the Failover Cluster online.
Since I had to re-join the Hyper-V hosts to the domain, there was something wrong with the relationship between one of the hosts and the Cluster Name Resource. So the Cluster still wouldn't come online.
This could however be fixed by right clicking the Resource and selecting "Repair Active Directory Object".
So after a very unpleasent night and morning, I got the cluster back online.
So I can get the remaining VMs migrated over to my New Cluster.
Maybe it sounds very straightforward, but there were many little problems that had to be solved underways :)
Thanks for the contributions.
R
I had to work continously through the night and on Friday before I managed to solve the issue(s).
To summarize:
I managed to remove the CSV from the cluster, and Access the volume from Windows Explorer on one of the Hyper-V hosts.
This allowed med to copy the VHD-file for the VM I needed (the DC) to the local disk and create a new VM based on this file.
After bringing the DC online, when I tried to log on to the Hyper-V hosts, I got an error with the Security/trust relationship between this computer and the domain.
Then I had to log on with local admin, remove the server from the AD domain, reboot, re-join the domain and reboot.
Not a big problem, except that one of the 2 servers didn't have the standard password for the local admin account.
This was around 4 AM, didn't make me very happy :)
When I finally got around this problem and had gotten both Hyper-V hosts re-joined to the domain, with the DC running, I could get on with getting the Failover Cluster online.
Since I had to re-join the Hyper-V hosts to the domain, there was something wrong with the relationship between one of the hosts and the Cluster Name Resource. So the Cluster still wouldn't come online.
This could however be fixed by right clicking the Resource and selecting "Repair Active Directory Object".
So after a very unpleasent night and morning, I got the cluster back online.
So I can get the remaining VMs migrated over to my New Cluster.
Maybe it sounds very straightforward, but there were many little problems that had to be solved underways :)
Thanks for the contributions.
R
FYI: We always have a physical DC in our cluster settings. It is time authority on the network and provides an outside AD/DNS for cluster management and ongoing operations.
As far as the password situation look up: Crack the Cred episode 105.
Dana gives a simple step-by-step for resetting that password so long as the OS volume is not BitLocker encrypted.
As far as the password situation look up: Crack the Cred episode 105.
Dana gives a simple step-by-step for resetting that password so long as the OS volume is not BitLocker encrypted.
Don't worry CSSGT. We've all been there, fixing a cluster that won't come online until the wee hours of the morning.
For what it's worth I also do the same thing as Philip; have a physical DC that's not part of the cluster whatsoever.
For what it's worth I also do the same thing as Philip; have a physical DC that's not part of the cluster whatsoever.
ASKER
Yes, always have a DC available not on the cluster :)
Problem was that this is a cluster that's part of an AD forest we're no longer a part of, after splitting up with another Company. But since you can't change the domain on the cluster to our new AD, it's still running until I get the last VMs out.
Not a preferable situation, as I was forcefully reminded of :)
R
Problem was that this is a cluster that's part of an AD forest we're no longer a part of, after splitting up with another Company. But since you can't change the domain on the cluster to our new AD, it's still running until I get the last VMs out.
Not a preferable situation, as I was forcefully reminded of :)
R
As an alternative one can run a DC directly on the cluster node. That is, set up the VM on local storage within the local Hyper-V management. Fellow cluster and Hyper-V MVPs have indicated they do this in the absence of a physical DC on at least two nodes in the cluster.
For the cost of a Windows Server license and an HP MicroServer if an inexpensive 1U is not purchased it's worth it IMNSHO to have a physical DC. :)
For the cost of a Windows Server license and an HP MicroServer if an inexpensive 1U is not purchased it's worth it IMNSHO to have a physical DC. :)
- Is the CSV on some sort of SAN or NAS device?
- How do the cluster nodes connect to the CSV device: iSCSI, SAS, fibre?
- In the Failover Cluster Manager console, who is the Owner of the CSV set to and is this node online?
- Do you have a DC that is not on the failover cluster?
- Have you tried running the Cluster Validation wizard?