I have spent some time troubleshooting an issue with an NSX 6.2 Controller deployment and have arrived at maybe the simplest (but not something I have found documented anywhere) solution.
Through obvious troubleshooting (read the error, read the logs, read the KBs) I tried a lot of steps (which IMO are worth trying if you are faced with the same error yourself):
- Ensure connectivity - run pings from the NSX manager to vCenter / Controller, etc. Make sure your network is flat in the management cluster and you're not doing unnecessary routing.
- Check the subnet mask is consistent all over. This has caught others out.
- There's a KB for clearing ARP cache on the physical switches (although this seems to be resolved from vCNS 5.something.
- Checked the accounts I was using to register NSX in vCenter were correct... Changed these, added permissions, etc...
The problem in the end - the password I was using was Passw0rd!"£$% (12345 with the shift key on a EN-GB keyboard pressed). Another of the above steps was to avoid using special characters you'll only find on a EN-GB keyboard, so I tried Passw0rd!Passw0rd!.... Still no joy.
I tried changing it to a different string that was:
- Above 12 characters
- Had the requirements (upper, lower, special (I used !) and number
- Didn't contain the word 'Passw0rd'
And it worked. The password I changed it to was very similar to password in length and use of a 0 - but not password..... And sure enough... I have managed to deploy the first controller (after 11 failed attempts!)