Hi All,
NSX-T 4.0 (later upgraded to 4.1) was installed via vcenter plugin. Now I can't add another vcenter as a compute manager, the error is
"Cannot add new compute manager, since NSX was create through VC driven installation. (Error code: 90017)"
But I could'nt find any information about this limitation. There wasn't any mention in the installation guide or the administation guide that it is not possible to use several machine managers if you installed nsx-t via vcenter plugin.
So what can I do to add another compute manager?
New deployment for me with ver. 4.1, and I too receive the same error. No idea how to resolve it yet, and no articles could be found referring to this message. I am working with support for deploying NSX for us, so hopefully they can find a solution. Not exactly a reassuring start to our NSX experience.
Hi,
which form factor did you use when deploying? Note that the medium form factor supports a maximum of 2 Compute Managers per NSX Management Cluster (VMware Configuration Maximum tool).
Have you already deployed 2 Compute Managers? The error code on the message indicates a limit has been exceeded...
Regards,
Lukas
Hi Lukas,
There is only one Compute Manager from which NSX-T was deployed. I've used "Large" deployment type.
Hi,
hm... I'm not aware of any other "hard" limits then...
Did you already set up a 3-node Management Cluster or are you running a single Manager setup? I remember it was a limit in NSX-V that a single manager only can manage a single vCenter, although this limit seems not to be present on the configmax website.
There could also be a difference when using the UI directly served by the manager and not via the vSphere plugin.
Yes, it's a 3-node cluster.
I've tried to add the second compute manager from the vCenter plug-in and directly from the nsx-t node with the same result.
Hi,
I had the chance to test this in my demo deployment.
Please check under System > System Overview in the Capacity tab how many Compute Managers are supported. In my demo it was deployed as Medium and i have a 2 system capacity.
I encountered the same error as you and could only add the second vCenter by deleting the VcLink Computer Manager. After that I added the original vCenter again and could add the second one.
Note that this won't work if you already have a existing Transport Node Configuration applied to hosts on this vCenter and after this you won't get the NSX plugin in the vCenter(s).
My best guess is that NSX is hard-coded and bound to the originating vCenter when deployed directly, based on the fact that in the default state the construct doesn't even use the built-in Service User function.
I'll open a case with VMware as this is unexpected behaviour and isn't documented properly.
Sorry that I can't help you any further.
Thanks for you try.
Interesting, but I don't see any limitations about Compute Managers in the "Capacity" tab.
It will be pain for us, because I've already deployed transport nodes and our firewall guys have published many DFW rules for microsegmentation...
Is your issue got resolved ?
Has support been able to help you with this issue yet? I'm experiencing the same issue with our deployment.
Hi everybody,
sorry for the late reply.
The support is still working with VMware engineering on it - but based on the response and the fact that it was forwarding to engineering and is worked on for over two weeks now, i think that this is a bug and should work normally...
I will keep you posted.
Hi everybody,
I finally got a final statement on this - let me quote directly:
Engineering have confirmed that NSX deployed directly from vCenter would indeed hold a 1:1 relationship with that vCenter, and it was not originally designed to host more than one vCenter.
However it is possible to detach the NSX instance from vCenter to allow for more than one Compute Manager instance and this can be done either via the way i've mentioned above or with the support team's help.
I mentioned that this limitation should be more visible in the official documentation. Let's hope it shows up there some time ![]()
The same issue here... the transport nodes are already configured. Is there any detailed documentation available for the remediation steps? or following steps are correct for the remediation?
1. Remove NSX from each host transport node
2. delete the vclink compute manager
3. add the original vcenter compute manager
4. install NSX to each transport node
Thanks,
Chan
Hi,
that's what I've done to get it to work, but this will only be useful in a demo environment or if you have freshly deployed the NSX instance. I have no further information as VMware Support specifically advised contacting them in this case.
