From the SP, use the "system console" command to connect to the console, then log into ONTAP. From there the process is the same as it would be over ssh.
7-mode (repeat on each controller):
halt -f
cDOT (halts the entire cluster):
halt * -inhibit-takeover true -skip-lif-migration-before-shutdown true -ignore-quorum-warnings true
That should leave the systems at the LOADER prompt. After the maintenance event if its still sitting at the loader:
boot_ontap
Repeat on each node, wait for things to settle, then run config advisor to check for anything unexpected.
If this post resolved your issue, help others by selecting ACCEPT AS SOLUTION or adding a KUDO.