@EH_GIP_SYSTEMS /David,
To begin with, the perfromance in the cluster does not depend on what Ip address the cluster components have. So you can change IP addresses for the cluster components and expect the same perfromance if the connecting devices like switches and routers are configured to handle the new IP addresses and vlan in the environment.
The best document I know, to perform the task is:
“https://portal.nutanix.com/#/page/docs/details?targetId=Advanced-Setup-Guide-AOS-v511:ipc-cvm-ip-address-reconfigure-t.html”
The above is for AOS 5.11 and the components associated like hypervisor and IPMI. You can get the same document for older releases as well, they are mostly similar as long as your AOS is not several years old (Anything above AOS .5.5 should work).
There is a script that does the job called ”external_ip_reconfig” that does the real change, but for this to work properly you need to follow all steps recommended there, as missing a necessary step can cause un-necessary delays in the process.
As it states in the document, there are three scenarios that we need to pay attention to:
- Change the IP addresses of the CVMs in the same subnet.
- Change the IP addresses of the CVMs to a new or different subnet.
In this scenario, the external IP address reconfiguration script works successfully if the new subnet is configured with the required switches and the CVMs can communicate with each other in the new subnet.
- Change the IP addresses of the CVMs to a new or different subnet if you are moving the cluster to a new physical location.
In this scenario, the external IP address reconfiguration script works successfully if the CVMs can still communicate with each other in the old subnet.
As you notice for the second and third scenarios, you probably need assistance of your network administrator (if it is not you), to configure the switches properly.
Review the above document several times and ask any questions about it. I will be happy to respond to your concerns; and If you have never done this before, I suggest you to open a case with our support center.
One last suggestion that could probably make any compilation in the process easier to handle, is the fact that as long as you have access to IPMI address of the hardware, we can modify any network configurations to make things work in the cluster. This is easiest when you have physical access to the hardware. You can access BIOS to change the IPMI vlan and network settings for each hyperviosr.
I realize the above is too much to go over, but I will be happy to respond to any concerns you have about it.
Hello@sbarab,
Thanks for your answer.
We have already looked the document you have shared, it will be the process we will apply and we were looking for some feedbacks of users who have already apply it.
Because in the document this sentence, made us think than it was not so easy and could have some consequences (not proper cleaning, etc) :
Before you decide to change the CVM, hypervisor host, and IPMI IP addresses, consider the possibility of incorporating the existing IP address schema into the new infrastructure by reconfiguring your routers and switches instead of Nutanix nodes and CVMs.
That’s why we are trying to understand if this changer can have an impact or not on the cluster. So if you confirm not, it’s ok for us and when we plan to perform the task we will open a support case before to start, to be sure than we can be supported if the things go wrong.
David.
@EH_GIP_SYSTEMS
As @sbarab quoted if you have never performed the procedure and have doubts regarding it and need assistance feel free to open a case and get expert guidance from Nutanix Support during the configuration change.
@EH_GIP_SYSTEMS /David,
I can confirm that cluster performance and operations may not be affected by simply chaining the IP address to another subnet or vlan. The determining factors (given the fact that the hypervisors and CVMs are configured correctly) are the new switches and routers involved, as they need to be configured properly so they can accept cluster network packets and route them to the correct destinations and mange the network traffic.
To get back to your point, if you have the option of the moving the involved switches and routers with the hypervisors and configure them in the new environment exactly the same as before (which may be a hard task to make sure the network devices can access all they need Internally and externally), then you need not change any thing on the cluster side.
However, I believe you may be better off just arranging for the change of cluster IP following the document mentioned and of course with help from our support as was mentioned by myself and @HITESH0801 .
Regards,