Hi, I have experienced "Stale NFS file handle" issue when switching between oldnfs and newnfs on a CURRENT box (NFS server exporting ZFS mountpoints). The cause was that fsid was changed in the following conditions and not in the NFS subsystem itself, but I am wondering if these are expected behavior...
First, I tried the following configurations of NFS and ZFS, and saw if fsid of the same mountpoint (a mounted ZFS dataset) changed or not by using statfs(2): compile opts kld module fsid[0:1] kld loaded by ---------------------------------------------------------------------------- NFSSERVER+NFSCLIENT zfs 865798fa:8346ef02 loader NFSSERVER+NFSCLIENT zfs 865798fa:8346ef07 kldload(8) NFSSERVER+NFSCLIENT+ NFSD+NFSCL zfs 865798fa:8346ef03 loader NFSSERVER+NFSCLIENT+ NFSD+NFSCL zfs 865798fa:8346ef08 kldload(8) NFSSERVER+NFSCLIENT nfsd+nfscl+zfs 865798fa:8346ef08 loader ---------------------------------------------------------------------------- The fsid seems to depend on the module loading order of ZFS. Even on a running system, the following procedure changed the fsid, for example: 1. Boot with a kernel with NFSSERVER+NFSCLIENT. 2. Load zfs.ko and opensolaris.ko by loader. 3. Check fsid of a ZFS mountpoint --> 865798fa:8346ef02 4. Unload zfs.ko and opensolaris.ko, and then do zfs umount -a. 5. Load zfs.ko and opensolaris.ko by kldload(8), and do zfs mount -a. 6. Check fsid of the same mountpoint --> 865798fa:8346ef08 Is this by design or not? While testing the above configurations, fsid of a UFS mountpoint was not changed (as expected). -- Hiroki
pgpwjCMvEskbj.pgp
Description: PGP signature
