> On Dec 13, 2024, at 5:28 AM, Edgar Fuß <e...@math.uni-bonn.de> wrote:
>
> May I re-ask my question:
>> When I take down an NFS server, update the base system (say, from -6 to -8),
>> and bring it up again, will clients that have NFS handles open across the
>> downtime/update access the same files (under the assumption that I mount
>> everything to the same place after the update) under their handles?
>
>> In other words: is the NFS handle to VFS whatever stable across NetBSD
>> versions?
> My impression is that ffs_vptofh() and {ffs,ufs}_fhtovp() haven't changed
> for ages, but is this sufficient for NFS handles staying stable across an
> update from -6 to -8?
It entirely depends on how an individual file system implements those methods,
which may also include “which dev_t does the device have”. If you plug in
another disk, and they suddenly enumerate in a different order, it’s entirely
possible they won’t be stable **across re-boots**.
-- thorpej
