Jakob Tewes wrote:> Hey folks,
>
> i?m trying my luck with scriptbased zfs replication and got no more ideas
left so here comes my layout.
>
> Got a small machine with two zfs pools, one protected via raidz2 and one
including just 1 disk. Now i wanted
> to use zfs?s nice snapshot/replication options to ship data from the
"unprotected" to the "protected" pool.
> Made snapshots of the included voulmes, shiped the snaps via "zfs
send | zfs receive" - everything worked
> as expected. The "shiped" volume copied including all filesystem
options held in that pool, also including
> the mountpoint. Now i got both zfs volumes with same mountpoint
option. Caused by the running processes
> accessing the originating volume (and after replication even the
target volume - it also gets zfs-mounted)
> im not able to change the mountpoint on the target volume nor unmount
it.>
> i?d be very thankful if somebody could help me with an idea about how to
avoid
> the target volume to get mountet/receive the mountpoint option without
interfering with the source volume.
>
What you really want is
6883722 want ''zfs recv -o prop=value'' to set initial property
values of
received dataset
Until that''s available, you could use the receive -u option to avoid
mounting the received dataset, or you could set canmount=noauto on the
source dataset before sending. If you set the mountpoint property
locally on the received dataset, subsequent incremental receives should
leave the mountpoint alone (after build 128).
Tom