3

I have a single physical host, running 2008 R2 with Hyper-V, all fully updated. On that host I am running 3 clients, a DC, a web server and a SQL Server 2008 R2 SP1. All are running on Windows Server 2008 R2, again all fully patched.

Generally all is fine, but sometimes and not repeatably, when I shutdown the host properly, which suspends the Clients, when it all comes up again, SQL Server is no longer running and the IIS App Pool I am running the website in needs to be told which account to run in again!

Any ideas!?

Improve this question
5
  • Are you running just that single DC? Do your IIS App Pools and SQL Server utilize AD Accounts?
    – Brent Pabst
    Jul 5, 2012 at 13:58
  • "SQL Server is no longer running" - is that the virtual machine? Or only the SQL Server software on a running VM?
    – TomTom
    Jul 5, 2012 at 14:00
  • @TomTom - the VM is running, it is SQL Server itself that is not.
    – Colin Desmond
    Jul 5, 2012 at 14:03
  • @Brent - Yes, it is a single DC. The AppPool and the SQL Server are running in AD defined accounts.
    – Colin Desmond
    Jul 5, 2012 at 14:04
  • 1
    That sort of explains it, does it not ? ;)
    – TomTom
    Jul 5, 2012 at 14:10

2 Answers 2

Reset to default
4

Colin,

This is most certainly a permissions issue. Even though the machines are saved they cannot contact the DC when they come back online. As a rule of thumb you can virtualize DCs but you should really follow MSFT's best practices as it is advisable to run a separate physical DC for failover reasons. For instance, using clustering and other services on the host machine that rely on a DC will all fail because the guest DC vm has not started.

Unfortunately if you are trying to handle recoveries from a failure there is no way to ensure proper automatic startup, where the machines boot in a specific order. I found a couple of solutions:

  1. Hyper-V Startup Delay to the rescue: http://mctexpert.blogspot.com/2011/02/how-to-configure-start-up-delays-for.html
  2. Boot them up manually. You can change the start-up type from auto or last state to off and then boot them once your host is up.
  3. You can also change the mode from saved to shutdown and then add a boot wait timer by configuring a startup boot option select/delay to delay your IIS and SQL boxes from booting by like 2-5 minutes to give the DC a head start.
  4. If you do ever get into clustering you can configure the HA settings to better control which machines boot when.
Improve this answer
4
  • Thanks Brent, I'll change the VMs to shutdown properly. Then on startup, boot them all, but delay the IIS and SQL Server using this technique windowshell.wordpress.com/2010/07/06/…
    – Colin Desmond
    Jul 5, 2012 at 14:17
  • Yea, we also got a similiar rec'd from Dell when using an iSCSI array with our clustered/HA Hyper-V instances. Lots of problems with the clustered disks not being brought online because our DCs were not yet available. It's fricking ugly, but it works.
    – Brent Pabst
    Jul 5, 2012 at 14:22
  • Actually, I've used the Hyper-V delay in your first point.
    – Colin Desmond
    Jul 5, 2012 at 14:48
  • Yea, it's a little cleaner. That option was not "I don't think" available a few years ago.
    – Brent Pabst
    Jul 5, 2012 at 14:50
1

Looks like a domain authentication problem. Make sure you configure Hyper-V to start up the DC before the other servers, or they will not be able to contact their domain when they start, and will experience all sort of authentication issues.

Improve this answer
2
  • When we boot the VMs, we boot DC first then the others. This is about resuming them when suspended. Can we ask Hyper-V to resume in a particular order?
    – Colin Desmond
    Jul 5, 2012 at 14:07
  • 2
    No, BUT: you an put in a start DELAY which does the same thing.
    – TomTom
    Jul 5, 2012 at 14:10

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .