Comment 28 for bug 1931507

I came to report the bug, saw that the fix had shipped, and can confirm that it works. Thanks to Marc (and anyone in the background) for the quick response!

Getting the NFS server running properly again requires either a reboot or a certain amount of tinkering; when rpcbind stops and restarts, it forgets all its rpc-program-to-IP-port mappings (they are stored only in memory). To get everything working again without rebooting, do

   systemctl stop nfs-server
   systemctl start nfs-server
   systemctl restart rpc-statd
   systemctl restart quotarpc

quotarpc matters only if you use disk quotas. If you don't make sure statd is running and registered with rpcbind, both on the server and on all clients, NFS locking may not work (and processes trying to make locks may hang); this is most noticeable on a mail server. restart nfs-server should work, but when I tried it it waited forever and when interrupted left NFS down; stop/start was quick and effective.