Post Power On Script Question

[esnmb]

In my vSphere 5 U2 environment I am running SRM 5.02 with SAN replication.

My question is about post power on scripting.  Since my VM's have to have their IP Addresses changed, I find that I need to manually log in to some systems and run an ipconfig /registerdns.

I was wondering if I can just put that in as a post power on script by simply putting in ipconfig /registerdns right in the the "Content" field of the post power on step.

Thanks.

[TheITHollow]

I believe you can do this from a PowerCLI script.  In 4.1 and higher you should have the ability to use the invoke-vmscript commandlet.  From there you can gain access to the ipconfig /registerdns command.

For more information, check out Alan Renouf's site below:

http://www.virtu-al.net/2010/02/05/powercli-changing-a-vm-ip-address-with-invoke-vmscript/

[basher]

To simply answer your question - yes, you can.

I'm curious, what is specific about these VMs, that require running this command? Are they set by DHCP?

Thanks

Stefan Tsonev

[esnmb]

No these all have static ip's.  But we are changing the ip addresses on the DR side, so I want to make sure the servers update DNS properly.

[basher]

What SRM does is it restarts the VM after IP customization. When Windows re-boots it should update DNS.

But I guess you can force this update using ipconfig /registerdns command in a call-out.

[esnmb]

Exactly, but we have seen it register just fine, then 10-15 minutes later we noticed several of the VM's drop out of DNS until we manually run /registerdns...  So this is more of a CYA type thing.

[basher]

Just throwing in some ideas:

Is this during test or during actual failover? Could it be that somehow the production and the test/recovered VM are modifying a shared DNS server? Or do you have DNS replication and the DNS server at the protected site overriding the updated DNS record on the recovery site?

Stefan Tsonev

[esnmb]

This was on an actual failover, which we did on 9 sandbox servers, so all "prod" systems were off and datastores disconnected from prod ESX hosts.  We have domain controllers in our DR site which all are AD integrated DNS.

[dphv]

No answer to your original question sorry, but we use DHCP reservations where servers really need "static" IP addresses, otherwise just plain old DHCP whereever possible.

I was finding the script to reconfigure IP addresses was getting affected by automated windows restart to apply the windows patches that had invariably applied just prior to wanting to failover.