[OmniOS-discuss] zfs send | recv

[priyadarshan]

Hi Oliver,

Thank you for sharing your use-case on Nexenta.

A few years ago we also needed to have a DR site, and I tested several third party tools.

The introduction of ZFS to Linux, with ZOL, made things more confused, by adding several more tools (like sanoid: https://github.com/jimsalterjrs/sanoid).

Ultimately we settled on a simple tool, which we tested over the years with satisfactory results (zfsnap).

Of course, if Nexenta will not allow shell access anymore, that makes it a bit more difficult.

I would defer to more expert people, like Guenther, who also replied to your message.

Kind regards,

Priyadarshan


> On 11 Jun 2018, at 10:07, Oliver Weinmann <oliver.weinmann at telespazio-vega.de> wrote:
> 
> Hi Priyadarshan,
> 
> Thanks a lot for the quick and comprehensive answer. I agree that using a third party tool might be helpful. When we started using the two ZFS systems, I really had a hard time testing a few third party tools. One of the biggest problems was that I wanted to be able to use the omnios systems as a DR site. However flipping the mirror always caused the nexenta system to crash. So until today there is no real solution to use the omnios system as a real DR site. This is due to different versions of zfs send and recv on the two systems and not related to using third party tools. I have tested zrep as it contains a DR mode and looked at znapzend but had not time to test it. We were told that the new version of nexenta no longer supports an ordinary way to sync snaps to a non nexenta system as there is no shell access anymore. Nexenta 5.x provides an API for this. I need to find some time to test it.
> 
> Best Regards,
> Oliver
> 
> 
> 
> 
> 
> Oliver Weinmann
> Head of Corporate ICT
> Telespazio VEGA Deutschland GmbH
> Europaplatz 5 - 64293 Darmstadt - Germany
> Ph: +49 (0)6151 8257 744 | Fax: +49 (0)6151 8257 799
> oliver.weinmann at telespazio-vega.de
> www.telespazio-vega.de
> Registered office/Sitz: Darmstadt, Register court/Registergericht: Darmstadt, HRB 89231; Managing Director/Geschäftsführer: Sigmar Keller-----Original Message-----
> From: priyadarshan <priyadarshan at scs.re>
> Sent: Montag, 11. Juni 2018 09:46
> To: Oliver Weinmann <oliver.weinmann at telespazio-vega.de>
> Cc: omnios-discuss at lists.omniti.com
> Subject: Re: [OmniOS-discuss] zfs send | recv
> 
> 
> 
>> On 11 Jun 2018, at 09:11, Oliver Weinmann <oliver.weinmann at telespazio-vega.de> wrote:
>> 
>> Hi,
>> 
>> We are replicating snapshots from a Nexenta system to an OmniOS system. Nexenta calls this feature autosync. While they say it is only 100% supported between nexenta systems, we managed to get it working with OmniOS too. It’s Not rocket science. But there is one big problem. In the autosync job on the Nexenta system one can specify how many snaps to keep local on the nexenta and how many to keep on the target system. Somehow we always have the same amount of snaps on both systems. Autosync always cleans all snaps on the dest that don’t exist on the source. I contacted nexenta support and they told me that this is due to different versions of zfs send and zfs recv. There should be a –K  flag, that instructs the destination to not destroy snapshots that don't exist on the source. Is such a flag available in OmniOS? I assume the flag is set on the sending side so that the receiving side has to understand it.
>> 
>> Best Regards,
>> Oliver
>> 
> 
> Hello,
> 
> OmniOS devs please correct me if mistaken, I believe OmniOS faithfully tracks zfs from illumos-gate.
> 
> One can follow various upstream merges here:
> https://github.com/omniosorg/illumos-omnios/pulls?q=is%3Apr+is%3Aclosed
> 
> Based on that, illumos man pages also apply to OmniOS: https://omnios.omniti.com/wiki.php/ManSections
> 
> Illumos zfs man page is here: https://illumos.org/man/1m/zfs
> 
> That page does not seem to offer a -K flag.
> 
> You may want to consider third party tools.
> 
> We have a very similar use-case as you detailed, fulfilled by using zfsnap, with reliable and consistent results.
> 
> git repository: https://github.com/zfsnap/zfsnap
> site: http://www.zfsnap.org/
> man page: http://www.zfsnap.org/zfsnap_manpage.html
> 
> With zfsnap we have been maintaining (almost) live replicas of mail servers, including snapshotting, either automatically synchronised to master, or kept aside for special needs.
> 
> One just needs to tweak a shell script (or simply, one or more cron jobs) to what is desired.
> 
> 
> Priyadarshan
> 
>