[zfs-discuss] zfs recv TXG Flushes

Aneurin Price aneurin.price at gmail.com
Thu May 26 05:57:32 EDT 2016


On 26 May 2016 at 00:15, Richard Laager via zfs-discuss
<zfs-discuss at list.zfsonlinux.org> wrote:
> I was reminded of this by the other thread about zfs recv.
>
> It feels like the speed of my receives on certain systems is limited by the
> sheer number of snapshots, not the amount of changed data.
>
> I haven't studied this in detail, but I get the impression that `zfs recv`
> has to do a TXG flush for each snapshot that it receives. Does anyone know
> if that's correct?

I can't answer that, but I can confirm that IME receiving a large
number of snapshots can take an extremely long time even if no data
changed in any of them, so it certainly sounds likely.


More information about the zfs-discuss mailing list