Exchange 2007 SP1 in Hyper-V
- I created my first Hyper-V server in a Dev Environment, but it was just a stand-alone desktop that supported Hardware Virtualization with a single IDE disk (only a single partitions. In installed Server Core 2008 w/ Hyper-V. Then installed Exchange 2007 SP1 (Mailbox Role) into a Guest VM on a single Fixed VHD. Migrated all my mailboxes, and used this Dev environment in production for about two weeks; and it worked GREAT.
Once my new Production Server was ready to go with teamed NIC's and Hyper-V installed; just the same as in the Dev environment, I migrated the VM/VHD from the Dev PC to the new Server. I should mention the server has one large RAID array across 8 HD's. Just like in the Dev environment, there is virtually only one disk. The Host OS is on the disk, and the Guest VHD's are just files in the "ProgramData\...\Hyper-V" folder on that disk.
Every few days the new Server Locks up. I cannot RDP to it, log onto the console, and Hyper-V Manager from a remote workstation cannot connect to the Hyper-V RPC service. The Guest VM's work for a while, but then also lock up. After restarting the server I see alot of Event ID 57 (VolMgr) and Event ID 51 (Disk) errors in the Host OS Event Logs; and little issues in the Guest VM's Event Logs.
The server restarts fine, without errors, and runs for a few more days, and then locks up again.
I can't seem to pin-down why this is happening, or what to look at. VSSAdmin of course after reboot does not report issues with any of the VSS Writers, and I can't get to looking at VSSAdmin prior to the hang/crash.
Where do I start here? How do we identify what is causing the VolMgr issues.
Directory of Technology
All Replies
- This sounds like a Hardware issue on the Host system. I would run some hardware diagnostics to see if anything shows up.
Thanks
Will
Will Shepherd - MCSE/MCITP/MCTS (Windows 2008,Exchange 2007,OCS 2007) - Check and update the drivers of NIC. Also run a diagnostics utility from the HW vendor. i have seen this issue and got resolved with drivers update on NIC.
Raj - Rajnish -
It's interesting that you suspect the NIC; because that is something 'unique' on this system. The server is a Dell R710 (brand new). Dell did/does not have animplementation suggestion for NIC Teaming between the Host OS and the Core Network on Server 2008 Core (no .NET). I 'ideally' wanted to Team all 4 GB NIC's back to the Core Network; and then setup one "virtual NIC" in Hyper-V bound only to this Virtual Team Interface NIC. I contacted BroadCOM (the NIC manufacture for Dell) and was directed to their cmd-line syntax for setting up the NIC Team with their latest drivers on Server 2008 Core. As a result I have the network now implemented as I wanted it.
Just before publishing this post, I finally installed the ethernet cables on two of the GB Interfaces on the system. While they are members of the team, they had not yet been terminated. In a normal stand-alone server, this is usually of no consequence, as the team setup is just "Load Balancing and Failover".
What is the dependancy between the NIC architecture and the Host OS disk operations, that could explain the relationship of this with my issue?
Directory of Technology - Yesterday at about 1pm, the Error 51's began again. The server had been running for 3 hours and 20 min. It was completely unresponsive at the console; but the VM's were as usual, running without error.
Today I installed the Dell SAS 6i RAID driver on the server, to see if that would improve the situation. This was Vol_Lic EE 2008, not Dell OEM media. We purchased the server without the OS. So to date, the only drivers installed beyond Windows Setup are the NIC drivers, and now this Storage Controller driver.
I also considered updating the firmware on the RAID card, but upon launching the FW batch file, the .exe that performs the update fail for incompatibility issues with x64 bit windows. So the firmware remains original from ship date.
I will report in a few days is this has helped stabilize things.
Directory of Technology - I have an idea, to aid in troubleshooting, is there a way to 'force' the event that is causing my error.
Event 51 states: An error was detected on device \Device\Harddisk0\DR0 during a paging operation.
Event 57 that directly follows states: The system failed to flush data to the transaction log. Corruption may occur.
Is there a command I can run to "flush data to the transaction log"?
Directory of Technology - I checked couple of suggestions on this, and they suggest (to my surprise), Its not HW, probabaly the OS or the application. http://www.hardwareanalysis.com/content/topic/12946/?o=80
Not sure if the link can provide a soluion, but there are lot of suggestions that might help you to resolve the issue.
Raj - I'm having this exact problem.. It started October 23rd. I have two virtual hosts a Dell PE2950 and a Dell R805.. Both have been running for nearly a year and both on the week of the 23rd Started this exact problem.. The PE2950 is running a PERC 5/E and the R805 is running a
PERCE 6/E.. Both have Dell MD1000 DAS connected to them with just a pair of mirror drives for the Host OS. I upgraded the R805 to Windows 2008 R2 including updating the Tools on All instances and it still has the same problem.. The R805 has this problem once a day and the PE2950 has it once a week but it definately started on Both Machines on the Week of October 23rd.. No changes have been made and no updates were installed prior to this problem appearing and i already swapped out one of the PERC Controllers with a brand new Controll but that has not made any difference.. Any suggestions?
the internet? whats the internet? - mental note.. Mine aren't running exchange so i do not think exchange is the problem..
the internet? whats the internet? HI,
Seem you problem more close to Server issues or hyper-v issues. So please repost your problem in below forums as well.
Resource
Windows Server 2008 R2 Hyper-V
Windows Server
regards
Chinthaka Shameera | MCITP: EA | MCSE: M | http://howtoexchange.wordpress.com/
