This issue is resolved in NSX-T Data Center 3.1.0 and above.
Workaround:
Any Manager node which does not have a nonconfig partition will need to be replaced.
If the NSX Manager orchestrating the upgrade is not impacted then the Manager node(s) impacted should be replaced one at a time.
Rerun the prechecks and once they succeed, proceed with the upgrade.
If the NSX Manager orchestrating the upgrade is impacted, please follow these recovery steps.
1) ssh to the orchestrator NSX Manager as admin user
2) Stop the upgrade process
> stop service install-upgrade
3) Elevate to root with 'st en' and delete the Upgrade Plan via API call
curl -v -k -u admin -H "Content-Type:application/json" -X DELETE https://NSX_MGR/api/v1/upgrade-mgmt/plan
4) Replace the orchestrator node
5) Upload the upgrade bundle again via the UI
6) Rerun prechecks and proceed with upgrade