Hi Rick, Rick Macklem <rmacklem_at_uoguelph.ca> wrote in <468764384.310026.1314219682612.JavaMail.root_at_erie.cs.uoguelph.ca>: rm> It sounds like people have agreed that this is a reasonable solution. rm> If hrs_at_ can confirm that testing shows it fixes the original problem rm> (the ZFS file handles don't change when it's loaded at different times), rm> I'll pass it along to re_at_. I am sorry for the delay, but I tried the patch on several boxes and it worked fine: [old (fixed array) patch] % lsvfs Filesystem Num Refs Flags -------------------------------- --- ----- --------------- ufs 2 3 oldnfs 15 0 network zfs 7 4 jail, delegated-administration nfs 6 1 network cd9660 1 0 read-only procfs 3 0 synthetic devfs 4 1 synthetic msdosfs 5 0 [new (hash-based) patch] Filesystem Num Refs Flags -------------------------------- --- ----- --------------- ufs 53 3 oldnfs 77 0 network zfs 222 4 jail, delegated-administration nfs 58 0 network cd9660 189 0 read-only procfs 2 0 synthetic devfs 113 1 synthetic msdosfs 50 0 [new patch, different loading order of kld modules] Filesystem Num Refs Flags -------------------------------- --- ----- --------------- ufs 53 3 zfs 222 4 jail, delegated-administration cd9660 189 0 read-only procfs 2 0 synthetic devfs 113 1 synthetic msdosfs 50 0 nfs 58 0 network Thanks a lot for the patch. I think it should be committed before 9.0R is released. Even for 8-STABLE this is useful but there is a problem that it will make an incomptibility of the fsid calculation between 8.N and 8.(N+1). What do you think about adding a loader tunable (something like vfs.fsidhash) to control this and making it disable by default? It would help sysadmins who will try a upgrade from 8.X to 9.X in the future, I think. -- Hiroki
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:17 UTC