[zfs-discuss] ZFS Send for continue

Zak Smith zak at computer.org
Wed Dec 14 01:16:03 EST 2016


On Tue, Dec 13, 2016 at 09:35:30AM -0800, Benjamin Smith via zfs-discuss wrote:
> On Sunday, December 4, 2016 8:07:50 AM PST you wrote:
> > Problem with "zfs send" is :
> > when a stream create between source and destination in destination create a
> > snapshot  , so when in middle connection lost,
> > for next send, show a error:
> > a snapshot is exist and it dont send again,
>
> Sounds like the following is happening:
>
> 1) You start to send a snapshot from server A to server B, where the snapshot
> is immediately piped to a receive process on B.
> 2) The flow is interrupted at the network level and the snapshot doesn't
> completely send from A to B. (EG: cable pulled out of the wall, flaky
> connection, whatever)
> 3) You retry the send as in #1 above and server B complains about the snapshot
> already existing?
>
> My guess is that the zfs receive process on B started in step 1 needs to be
> stopped/killed. When I do this, I'm able to restart the send with the same
> snapshot without issue.

I do this a different way.  I always check for a common ancestor and
then send the snapshot diff from the most recent common snapshot to
the latest snap, and I have never had this have a problem when
re-trying during the next iteration, if the prior one was aborted due
to a network error




--
Zak Smith
307-543-7820 office
Please do not send private or confidential information via email.





More information about the zfs-discuss mailing list