How to HA Grid migration from Appliance to vNIOS?

[#1]

We are going to migrate our existing environment with new virtual ones.

Currently our grid exists of the following:

Old:

Master 2x IB-1410 (physical) (HA)

Member 2x IB-1410 (physical) (HA)

New:

Master 2x IB-VM-1415 (HA)

Member 2x IB-VM-1415 (HA)

New and old are all running software version: 8.3.6-385000

Requirement: all VIP IP's, hostnames and data should be maintained.

Replacing the master should not be a big problem i think:

- add HA pair to the grid

- make it candidate master

- let it sync DB

- promote it to master using cli.

- Remove old master from grid

- adjust hostnames and VIP IP

and we are done

But how do I accomplish the same with members?

I cannot make it candidate master since it will not be master at the end so what is the best approach to migrate and keep all data?

The new VM's are already in HA mode but not sure if that was a wise decision?

[#2]

Hi Richard,

You are correct. The steps you had mentioned for swapping the grid master should work.

To swap the existing HA pair, first you can perform a database reset on the passive node of HA member using the command "reset database" from CLI. Now join one of the new appliance to the existing grid as a passive node.

After syncing it with the existing grid, perform a failover so that the newly swapped passive node becomes the active node.

Once completed, you can perform a database reset on the new passive node which is an old appliance and replace it with the new appliance.

I hope it helps.

Thanks,

Alten Alexander

[#3]

Hi Alten,

Thanks for your explanantion but i thought you cannot replace hardware with virtual in HA (or vice-versa) in a cluster? Member type vNIOS versus Infoblox conflict plus 1410 vs 1415 ?

Let's say you can. With a bit more detail the steps would be like this then:

- login to PASSIVE node

- reset database from CLI

- swap PASSIVE node in HA config with NEW node (all devices are in the same subnet)

- sync database

- force failover so newly added node will be ACTIVE

- repeat above steps for current PASSIVE node

Did i understand it correclty?

[#4]

I wonder if there is any update on this. I recently started a P2V migration and so far it has been quite painful and the migration is currently on hold.

A single grid member in build site was replaced without issue. Doing the HA Grid Master Candidate was problematic.

The approach was to swap the passive node of the HA pair failover and then repeat. However when testing in the lab we saw that changing the "Hardware Type" was an issue. The incorrect Hardware Type "Infoblox" or "Virtual NIOS" cause the Join Grid to fail.

It is slightly unclear whether a full shutdown of the HA pair was required to be able to change the "Hardware Type" or what seemed to work,shutting down the passive node, attempting to join the new vNIOS passive node, this fails but after the failure, it seems the "Hardware Type" drop down becomes available.

We did have able 15 minutes of the GM being offline. So any experience anyone has had would be interesting to hear.

[#5]

Where you able to get this to work? Thanks

[#6]

Hello,  

Does anyone have a good documented process that can be shared for this.  I am trying to replace EOL appliances with new VMs.  The appliances are in HA and the VM will be single. 

Finding this process to not be documented well.  Want to preserve the HA  VIP for the new VM as the IP helpers on our Cisco switches point to the the HA VIPs.  

Not sure how to use the VIP for the new VM and preserve network memberships so I don't have to change either IP helpers or modify every network to update member properties.  

Any suggestions or be pointed to documents detailing this would be much appreciated.  

Thank you.  

[#7]

You can't mix and match physical and virtual members in an HA pair, they both either have to be physical or virtual. This capability is coming though, I believe it is being introduced in NIOS 8.6.

If you can't wait that long then what I would do is build a new virtual appliance as a grid master candidate on a new IP address, let it sync the database, then promote it to become the grid master. Once it is active you can then shut down the old physical HA pair. If you need to retain the IP address of the old grid master, then simply change the IP address of the virtual appliance to the old IP address. Once you have got it up and running you can then add a second virtual appliance to become it's passive HA node.

Now you have the grid master running virtually you'll need to convert the member HA pair. Again you can't mix physical and virtual, so you will effectively have to shut the pair down, change the hardware type in the UI to "Virtual NIOS", then join the virtual members to the grid using the same IPs. They will then sync with the grid master and pick up all the data that was on the old appliances. However you will have an outage whilst you do this, so worth doing this during scheduled downtime. You can do just do one node to start with to get the service up as quickly as possible then join the second virtual node in as the passive HA node later. So you'll probably need about 30 mins of downtime to do the switchover.

[#8]

Hello Richard,

Did the procedure you mentioned here worked ? we are planning to migrate from 1410 to V1415.