We've got a fairly solid process for re-imaging NFS servers now that was set up for T266202: Move or recable labstore1004 to 10Gbps rack (if needed) and ethernet.
That means we can move to buster as soon as we have:
- a buster build of ldap-nfsd https://gerrit.wikimedia.org/r/admin/repos/operations/debs/nfsd-ldap
- a test of DRBD replication between stretch and buster (significant DRBD upgrade there)
After that, it's a matter of re-imaging
We could potentially try to fix the naming on labstore100x, though that is likely to be a lot of trouble for very little benefit. I think it is better we just let those old hosts get refreshed away instead of coupling things like that.