You should just be able to click the link on that node (can't remember the title) and it will re-poll the node and attempt the VIB push. One of the leading causes of VIB installation failure is due to lack of space in the bootbank. NSX-T pushes out A LOT of VIBs and needs quite a good bit of room on the ESXi host.
Thanks but that's def not the issue here - I have ~16-32GB on the ESXi installation disk and they're greenfield builds. Something else got wonky, but no amount of clicking the URL in the status does anything. It offers to "remediate" or "resolve" but nothing happens. I've had to go and get the NSX manager thumbprint, then nsx cli on the host, detach, stop svcs, etc. and do "del nsx". But, that only worked on hosts that were install failed. Other ones show as NSX not configured/uninstall failed, yet there are no vibs on the host at all... because I rebuilt said host but NSX manager never like... re-checked. Or, if it does, it considers it the same host by name and nothing else and it's got stale stuff in its database.
I ran into a similair issue when i entered the wrong segment for the vmkernel migration.
I was able to remove the failed host using the api.
https://<NSX manager IP>/api/v1/transport-nodes/state
https://<NSX manager IP>/api/v1/transport-nodes/58293063-8aea-4a86-a67a-e285fc058eba?unprepare_host=false
Does this problem visible only with nested hosts?
Yes, in my case. I have seen similair issues on bare-metal hosts that could be resolved from the gui.
But i'm still not sure if the creator of this thread had the same issue.