Udostępnij za pośrednictwem

Script to evacuate cluster node on shutdown

  • Artykuł

A little while ago David posted a really handy solution for evacuating a Hyper-V cluster node on shutdown automatically.

The issue that he is trying to address is that normally you should not just shutdown a node in a cluster, rather you should manually move all virtual machines off of the cluster node first.  If you just shutdown a node in a cluster all the virtual machines will be failed over using quick migration.

David solved this problem by writing a PowerShell script that works with clustering to live migrate the virtual machines to another node in the cluster.  He then uses group policy to get this script to be run automatically whenever a computer in his cluster is shutdown.

All-in-all it is a really cool solution to make your life a little easier.  You can read all about it here:

https://pdconsec.net/blogs/davidr/archive/2010/09/01/Automatically-Live-Migrate-VMs-On-Host-Shutdown.aspx

Cheers,
Ben

Comments

  • Anonymous
    October 25, 2010
    What happens when a shutdown force command is executed? ie: shutdown /f Does the script still run?

  • Anonymous
    October 26, 2010
    I would assume not - but I need to confirm that.

  • Anonymous
    October 28, 2010
    Yes, the shutdown script will still run if you force the host to reboot with shutdown /f - local policy and group policy shutdown scripts are executed during the shutdown sequence. The original driver for the script was so that patching of the host servers with SCCM could occur automatically during maintenance windows; the reboots occur after patching is complete. There's one server per maintenance window - so at the end of the process, all servers have roughly the same number of VMs and one server is "empty". Then, SCOM/SCVMM Pro Tips take over and balance out anything grievously bad. Thanks for the mention Ben :)

  • Anonymous
    November 01, 2010
    David - Thanks for the extra info! Cheers, Ben

  • Anonymous
    November 02, 2010
    If you have a 2 node cluster, you get an error from trying to index an array when a single object is returned for available nodes.  Casting the statement as an array fixes this: $AvailNodes = @((get-clusternode | Where-Object { $.State -eq "Up" -and $.Name -ne "$Computer" }))

  • Anonymous
    July 12, 2012
    Since Automatic Start Action does not persist after a Live Migration I modified the previously provided code (currently not available), by using pshyperv.codeplex.com/.../38769  and adding at the end of script: Invoke-command { import-module C:TasksAutomaticStartActionHyperV.psd1 } -session $session Invoke-command { set-vm vm1 -autostart 2 } -session $session Invoke-command { set-vm vm2 -autostart 2 } -session $session

  • Anonymous
    July 12, 2012
    Correction: $session = New-PSSession -computerName hypervnode1 Invoke-command { import-module C:TasksAutomaticStartActionHyperV.psd1 } -session $session Invoke-command { set-vm vm1 -autostart 2 } -session $session Invoke-command { set-vm vm2 -autostart 2 } -session $session

  • Anonymous
    August 16, 2012
    Looks like this link is dead. Anyone still have the article or script?