We have a number of customers who wanted to layout their new cDOT infrastructure without using the ADP function, primarily to reduce their data aggregate count, and maintain consistency with other existing deployments. Typically these solutions comprise multiple shelves, so there are partitioned and non-partitioned disks to start with.
Start the build process as normal to establish the cluster, create a new small aggregate from the non-partitioned disks on each controller, move the root aggregates, remove and re-assign partition ownership so that any given disk and all of its partitions are owned by a single node, then unpartition the disks and zero the spares.
To move the root aggregate, refer to the following: https://kb.netapp.com/support/index?page=content&id=3013873&locale=en_US
To invoke the environment variable change use: setenv root-uses-shared-disks? false the first time you drop to Loader during the root aggregate move process.