Replication, re-ip and DNS updates

[Cragdoo]

Has anyone encountered this niggle with replicated VMs?

I have about a dozen VMs (inc 2 domain controllers) replicating to our hosted DR site, and the VMs all successfully re-ip, but when I carry out a failover test, the DNS on the replicated VMs still keeps the old IP address, and does not force a DNS refresh.

Is there anyway to force the DNS refresh at the DR site, when powering up the replica VMS ?

[Dima P.]

Hello Craig,

Just to clarify, if you flush dns (for example thru cmd) on the failed over machine the needed DNS settings are applied correctly?

[dellock6]

Wait, are we talking about dns configuration in the ip stack, or the dns resolver cache?

[Cragdoo]

dns resolver cache, let me try to explain in a little more detail

Production Subnet = 192.168.1.0/24
DC1 - prodution IP = 192.168.1.10
VM1- production IP = 192.168.1.33

DR Subnet = 172.20.2.0/24
DC1 - DR IP = 172.20.2.10
VM1 - DR IP = 172.20.2.33

I then perform a failover from the VBR console , VM1 has the re-ip applied and the DNS is set to the DC1 DR IP (172.20.2.10) , but DNS name of VM1 still resolves to the Production subnet 192.168.1.33 , how can I force the replicated VMs to re-register their updated IPs with DC1, with the new DR IPs ?

[dellock6]

The two commands you need to run in the failover vm are:
ipconfig/flushdns
ipconfig/registerdns
You cannot have however scripts in a replica job, but you can if you put the VM inside a failover plan, maybe this can be a good way to implement the commands.

For the future, if this behaviour is consistent among windows vms, we may think about adding these two commands as part of the re-ip process.

[Cragdoo]

Thanks Luca , I knew the commands , and I am using failover plans. So I just setup simple batch file, with

ipconfig /flushdns
ipconfig /registerdns

and set them as a 'Post-Failover script' in the failover plan?

This will run the batch against each VM in the failover plan ?

2 of the VMs are Linux, any suggestions how to handle DNS from this perspective ?

[dellock6]

The script is executed from the veeam console, so it has to connect to the VM(s) and do the actions.
On linux, I'd check first if it needs to force the refresh of the dns resolver cache or not. If yes, it depends on which resolver is in use, have a look for example here:
http://www.cyberciti.biz/faq/rhel-debia ... dns-cache/

[Cragdoo]

I'd have question the reason behind not putting the flushdns/registerdns into the re-ip process, updating DNS would have made sense, seeing as IPs are being changed ??

[Gostev]

Well, re-IP is done on an offline VM (by mounting its disks prior to powering it on). While executing the above commands obviously can be done on a running VM only. That said, I am confused as to why those commands are needed in the first place, when we are essentially doing cold VM boot during failover, so this activity (DNS registration) should happen during OS boot up anyway? May be something to investigate with our support.

[nunciate]

What are you using for DNS Servers and how are they configured? We use DNS on our 2012 AD controllers. We have DNS servers Pimary/Secondary here in Prod and 1 Tertiary in DR. They are configured to replicate to each other so that when a VM comes on it does an auto-update to DNS and registered the DR IP automatically. Haven't had any issues with that. I am assuming you have the Register this connections address address in DNS setting checked on all of your VM NICs and that your DNS is configured to allow automatic registrations.