RAID-[56] stabilisation

RAID-[56] stabilisation

[Roy Sigurd Karlsbakk]

Hi all

It's been a wee bit more than a decade since I started to look into btrfs as a possible ZFS replacement. I just wonder when one can expect RAID-[56] to stabilise as in usable for production, since this is, for my use, rather critial.

Vennlig hilsen / Best regards

roy
--
Roy Sigurd Karlsbakk
(+47) 98013356
http://blogg.karlsbakk.net/
GPG Public key: http://karlsbakk.net/roysigurdkarlsbakk.pubkey.txt
--
I all pedagogikk er det essensielt at pensum presenteres intelligibelt. Det er et elementært imperativ for alle pedagoger å unngå eksessiv anvendelse av idiomer med xenotyp etymologi. I de fleste tilfeller eksisterer adekvate og relevante synonymer på norsk.

Re: RAID-[56] stabilisation

[Qu Wenruo]

On 2022/4/20 07:02, Roy Sigurd Karlsbakk wrote:
> Hi all
>
> It's been a wee bit more than a decade since I started to look into btrfs as a possible ZFS replacement. I just wonder when one can expect RAID-[56] to stabilise as in usable for production, since this is, for my use, rather critial.

Currently the main problem for btrfs RAID56 is write-hole.

To solve it, either we need a big feature in extent allocator (to avoid
allocate extents in the partial written stripe), or go traditional
journal based RAID56 (needs a new on-disk format change, and new code).


If your use case can rule out power loss (by using UPS) and unexpected
shutdown (no way to prevent kernel panic though), OR you can afford a
full btrfs scrub before any write (can be very time consuming), you can
try btrfs RAID56.

There is some plan to implement a RAID56 journal, but there is no
concrete time table for that yet.


Or, you can run single device btrfs on md/dm-raid56.

Thanks,
Qu

>
> Vennlig hilsen / Best regards
>
> roy
> --
> Roy Sigurd Karlsbakk
> (+47) 98013356
> http://blogg.karlsbakk.net/
> GPG Public key: http://karlsbakk.net/roysigurdkarlsbakk.pubkey.txt
> --
> I all pedagogikk er det essensielt at pensum presenteres intelligibelt. Det er et elementært imperativ for alle pedagoger å unngå eksessiv anvendelse av idiomer med xenotyp etymologi. I de fleste tilfeller eksisterer adekvate og relevante synonymer på norsk.

Re: RAID-[56] stabilisation

[Qu Wenruo]

On 2022/4/20 10:17, Qu Wenruo wrote:
>
>
> On 2022/4/20 07:02, Roy Sigurd Karlsbakk wrote:
>> Hi all
>>
>> It's been a wee bit more than a decade since I started to look into
>> btrfs as a possible ZFS replacement. I just wonder when one can expect
>> RAID-[56] to stabilise as in usable for production, since this is, for
>> my use, rather critial.
>
> Currently the main problem for btrfs RAID56 is write-hole.
>
> To solve it, either we need a big feature in extent allocator (to avoid
> allocate extents in the partial written stripe), or go traditional
> journal based RAID56 (needs a new on-disk format change, and new code).
>
>
> If your use case can rule out power loss (by using UPS) and unexpected
> shutdown (no way to prevent kernel panic though), OR you can afford a
> full btrfs scrub before any write (can be very time consuming),

I mean "before any write after an unexpected shutdown/crash"

If you can ensure your fs is shutdown cleanly, then no problem at all.

> you can
> try btrfs RAID56.
>
> There is some plan to implement a RAID56 journal, but there is no
> concrete time table for that yet.
>
>
> Or, you can run single device btrfs on md/dm-raid56.
>
> Thanks,
> Qu
>
>>
>> Vennlig hilsen / Best regards
>>
>> roy
>> --
>> Roy Sigurd Karlsbakk
>> (+47) 98013356
>> http://blogg.karlsbakk.net/
>> GPG Public key: http://karlsbakk.net/roysigurdkarlsbakk.pubkey.txt
>> --
>> I all pedagogikk er det essensielt at pensum presenteres
>> intelligibelt. Det er et elementært imperativ for alle pedagoger å
>> unngå eksessiv anvendelse av idiomer med xenotyp etymologi. I de
>> fleste tilfeller eksisterer adekvate og relevante synonymer på norsk.