Incompat features: raid56 ... when creating a RAID6?

Incompat features: raid56 ... when creating a RAID6?

[Andreas Grosse]

Hello everyone!

I just wanted to create a RAID6 and got the following output:

> # mkfs.btrfs -d raid6 -m raid6 -L slowPool /dev/sd[cdefgh]
> btrfs-progs v4.4.1
> See http://btrfs.wiki.kernel.org for more information.
> 
> Label:              slowPool
> UUID:               85ddf6a7-51f1-4cc0-b6cd-d472277b0e86
> Node size:          16384
> Sector size:        4096
> Filesystem size:    27.29TiB
> 
> Block group profiles:
>   Data:             RAID6             4.01GiB
>   Metadata:         RAID6             4.01GiB
>   System:           RAID6            14.50MiB
> 
> SSD detected:       no
> Incompat features:  extref, raid56, skinny-metadata
> Number of devices:  6
> 
> Devices:
>    ID        SIZE  PATH
>    
>     1     4.55TiB  /dev/sdc
>     2     4.55TiB  /dev/sdd
>     3     4.55TiB  /dev/sde
>     4     4.55TiB  /dev/sdf
>     5     4.55TiB  /dev/sdg
>     6     4.55TiB  /dev/sdh\0

And then the line saying "Incompat features: ... raid56" came to my eyes. 
Reading the corresponding manpage, it says:
> raid56
>     extended format for RAID5/6, also enabled if raid5 or raid6 block groups
>     are selected\0

So why is raid56 marked as incompatible if I just created a file system with 
multiple disks using the RAID6 profile? Have I misunderstood something there? 
I am confused. Can somebody here lighten this up?

(If important: Gentoo Linux with kernel 4.5.0, btrfs-progs v4.4.1)

Andiÿôèº{.nÇ+‰·Ÿ®‰­†+%ŠËÿ±éݶ\x17¥Šwÿº{.nÇ+‰·¥Š{±ý»k~ÏâžØ^n‡r¡ö¦zË\x1aëh™¨è­Ú&£ûàz¿äz¹Þ—ú+€Ê+zf£¢·hšˆ§~†­†Ûiÿÿïêÿ‘êçz_è®\x0fæj:+v‰¨þ)ߣøm

Re: Incompat features: raid56 ... when creating a RAID6?

[Hugo Mills]

[-- Attachment #1: Type: text/plain, Size: 1526 bytes --]

On Tue, Mar 15, 2016 at 10:42:29PM +0100, Andreas Grosse wrote:
> Hello everyone!
> 
> I just wanted to create a RAID6 and got the following output:
> 
> > # mkfs.btrfs -d raid6 -m raid6 -L slowPool /dev/sd[cdefgh]
[snip]
> > Incompat features:  extref, raid56, skinny-metadata
[snip]
> And then the line saying "Incompat features: ... raid56" came to my eyes. 
> Reading the corresponding manpage, it says:
> > raid56
> >     extended format for RAID5/6, also enabled if raid5 or raid6 block groups
> >     are selected
> So why is raid56 marked as incompatible if I just created a file system with 
> multiple disks using the RAID6 profile? Have I misunderstood something there? 
> I am confused. Can somebody here lighten this up?

   It's a safety thing.

   The incompat flags are markers set in the filesystem to indicate
which features that particular FS uses. Each kernel version has a list
of features it can handle, and if it's asked to mount a filesystem
with a feature that it doesn't recognise, it'll refuse to do so.

   So, you've created a filesystem with the RAID5/6 feature, it's
marked as such in the FS (with the incompat flag "raid56"), and
attempting to mount that FS on a kernel that doesn't know about parity
RAID (earlier than 3.14, IIRC) will fail safely because the kernel
can't handle it.

   Hugo.

-- 
Hugo Mills             | Alert status upwards vermilion: High probability of
hugo@... carfax.org.uk | flash photography. Avoid wearing brogues.
http://carfax.org.uk/  |
PGP: E2AB1DE4          |

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: Incompat features: raid56 ... when creating a RAID6?

[Anand Jain]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=windows-1252; format=flowed, Size: 2203 bytes --]



On 03/16/2016 05:42 AM, Andreas Grosse wrote:
> Hello everyone!
>
> I just wanted to create a RAID6 and got the following output:
>
>> # mkfs.btrfs -d raid6 -m raid6 -L slowPool /dev/sd[cdefgh]
>> btrfs-progs v4.4.1
>> See http://btrfs.wiki.kernel.org for more information.
>>
>> Label:              slowPool
>> UUID:               85ddf6a7-51f1-4cc0-b6cd-d472277b0e86
>> Node size:          16384
>> Sector size:        4096
>> Filesystem size:    27.29TiB
>>
>> Block group profiles:
>>    Data:             RAID6             4.01GiB
>>    Metadata:         RAID6             4.01GiB
>>    System:           RAID6            14.50MiB
>>
>> SSD detected:       no
>> Incompat features:  extref, raid56, skinny-metadata
>> Number of devices:  6
>>
>> Devices:
>>     ID        SIZE  PATH
>>
>>      1     4.55TiB  /dev/sdc
>>      2     4.55TiB  /dev/sdd
>>      3     4.55TiB  /dev/sde
>>      4     4.55TiB  /dev/sdf
>>      5     4.55TiB  /dev/sdg
>>      6     4.55TiB  /dev/sdh
>
> And then the line saying "Incompat features: ... raid56" came to my eyes.
> Reading the corresponding manpage, it says:
>> raid56
>>      extended format for RAID5/6, also enabled if raid5 or raid6 block groups
>>      are selected
>
> So why is raid56 marked as incompatible if I just created a file system with
> multiple disks using the RAID6 profile? Have I misunderstood something there?
> I am confused. Can somebody here lighten this up?

  Those messages are indeed confusing. It just indicates the
  FS may fail to mount on certain older kernels. Not necessarily
  the kernel in the system on which you ran btrfs-progs. Sorry
  that it not very obvious at the moment, but there are patches
  to make this part better.

  To check features that your running kernel supports you could use
   ls /sys/fs/btrfs/features

  But note: Some of the names used by sys/fs/..features don't
  exactly match with the names used by btrfs-progs: incompatible..
  (a bug, which is also fixed in the patch).

Thanks, Anand


> (If important: Gentoo Linux with kernel 4.5.0, btrfs-progs v4.4.1)
>
> AndiN‹§²æìr¸›yúèšØb²X¬¶Ç§vØ^–)Þº{.nÇ+‰·¥Š{±nÚß²)í…æèw*\x1fjg¬±¨\x1e¶‰šŽŠÝ¢j/êäz¹Þ–Šà2ŠÞ™¨è­Ú&¢)ß¡«a¶Ú\x7fþø\x1e®G«éh®\x0fæj:+v‰¨Šwè†Ù¥
>

Re: Incompat features: raid56 ... when creating a RAID6?

[Andreas Grosse]

Am Mittwoch, 16. März 2016, 06:19:40 CET schrieben Sie:
>   Those messages are indeed confusing. It just indicates the
>   FS may fail to mount on certain older kernels. Not necessarily
>   the kernel in the system on which you ran btrfs-progs. Sorry
>   that it not very obvious at the moment, but there are patches
>   to make this part better.
> 
>   To check features that your running kernel supports you could use
>    ls /sys/fs/btrfs/features
> 
>   But note: Some of the names used by sys/fs/..features don't
>   exactly match with the names used by btrfs-progs: incompatible..
>   (a bug, which is also fixed in the patch).
> 
> Thanks, Anand

Hello Anand,

thanks for the detailed reply. Good to hear, those things are already patched. 
I thought that I could not be the only one who misunderstood "Incompat".
Thanks for the hint with the sys-files!

Andi

One more question: Any idea what happened here? Did I send this garbage? This 
was at the very end of your response mail...
> > (If important: Gentoo Linux with kernel 4.5.0, btrfs-progs v4.4.1)
> > 
> > AndiN‹§²æìr¸›yúèšØb²X¬¶Ç§vØ^–)Þº{.nÇ+‰·¥Š{±nÚß²)í…æèw*\x1fjg¬±¨\x1e¶‰šŽŠÝ¢j/êäz
> > ¹Þ–Šà2ŠÞ™¨è­Ú&¢)ß¡«a¶Ú\x7fþø\x1e®G«éh®\x0fæj:+v‰¨Šwè†Ù¥

Re: Incompat features: raid56 ... when creating a RAID6?

[Anand Jain]

> One more question: Any idea what happened here? Did I send this garbage? This
> was at the very end of your response mail...

  Yes looks like. As it was there when I read your posting.

>>> AndiN‹§²æìr¸›yúèšØb²X¬¶Ç§vØ^–)Þº{.nÇ+‰·¥Š{±nÚß²)í…æèw*\x1fjg¬±¨\x1e¶‰šŽŠÝ¢j/êäz
>>> ¹Þ–Šà2ŠÞ™¨è­Ú&¢)ß¡«a¶Ú\x7fþø\x1e®G«éh®\x0fæj:+v‰¨Šwè†Ù¥

Re: Incompat features: raid56 ... when creating a RAID6?

[Andreas Grosse]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset="UTF-8", Size: 649 bytes --]

Am Mittwoch, 16. März 2016, 08:52:52 CET schrieb Anand Jain:
>   Yes looks like. As it was there when I read your posting.
> 
> 
> >>> AndiN‹§²æìr¸›yúèšØb²X¬¶Ç§vØ^–)Þº{.nÇ+‰·¥Š{±nÚß²)í…æèw*\x1fjg¬±¨\x1e¶‰šŽŠÝ¢j/ê
> >>> äz
> >>> ¹Þ–Šà2ŠÞ™¨è­Ú&¢)ß¡«a¶Ú\x7fþø\x1e®G«éh®\x0fæj:+v‰¨Šwè†Ù¥

I'll have a look at this. I have no idea where this came from. But it seems 
this stuff is invisible on the list archives on the web on my initial 
message...

ÿôèº{.nÇ+‰·Ÿ®‰­†+%ŠËÿ±éݶ\x17¥Šwÿº{.nÇ+‰·¥Š{±ý»k~ÏâžØ^n‡r¡ö¦zË\x1aëh™¨è­Ú&£ûàz¿äz¹Þ—ú+€Ê+zf£¢·hšˆ§~†­†Ûiÿÿïêÿ‘êçz_è®\x0fæj:+v‰¨þ)ߣøm

Re: Incompat features: raid56 ... when creating a RAID6?

[Duncan]

Andreas Grosse posted on Wed, 16 Mar 2016 10:12:47 +0100 as excerpted:

> Am Mittwoch, 16. MÀrz 2016, 08:52:52 CET schrieb Anand Jain:
>>   Yes looks like. As it was there when I read your posting.
>> 
>> 
>> >>> AndiN‹§²Êìrž›yúÚšØb²X¬¶Ç§vØ^–)Þº
{.nÇ+‰·¥Š{±nÚß²)í ÊÚw*\x1fjg¬±š\x1e¶‰šŜŠÝ¢j/
ê
>> >>> Àz
>> >>> ¹Þ–Šà2ŠÞ™šÚ­Ú&¢)ßÂ
¡Â«a¶Ú\x7fßÞ\x1e®G«éh®\x0fÊj:+v‰šŠwÚ†Ù¥
> 
> I'll have a look at this. I have no idea where this came from. But it
> seems this stuff is invisible on the list archives on the web on my
> initial message...
> 
> N‹§²æìrž›yúèšØb²X¬¶Ç§vØ^–)Þº{.nÇ+‰·¥Š{±nÚß²)í…æèw*\x1fjg¬
±š\x1e¶‰šŽŠÝ¢j/êäz¹Þ–Šà2ŠÞ™šè­Ú&¢)ß¡«a¶Ú\x7fþø\x1e®G«éh®\x0fæj:
+v‰šŠwè†Ù¥


FWIW, your posting as it appeared on gmane's news (nntp, which I use for 
this list) and web archives was truncated.  Hugo evidently got a better 
copy than gmane did, as he quoted text that wasn't there for me.

Here's (one of) the gmane web link(s) to your post on the web archive:

http://permalink.gmane.org/gmane.comp.file-systems.btrfs/54108

As you can see, it's truncated there, tho the web interface doesn't let 
you see the raw message.

The nntp/news archive does better at letting you at the raw message.  I 
use pan as my nntp client, and fetched the raw message file itself from 
cache after getting the messageID from the pan GUI.  Unfortunately, not a 
lot of web or mail clients know what to do with links like...

nntp://news.gmane.org/gmane.comp.file-systems.btrfs

... and I don't remember how to add the specific article to the link.  So 
you may have trouble getting the article from news.

Tho AFAIK lynx, the text-mode browser, can handle news.  You might try 
that.


I could send you personally a copy of the message file, however, if you 
like.  Selected headers say:

User-Agent: KMail/ (Linux/4.5.0-gentoo; KDE/5.20.0; x86_64; ; )
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: base64


With the base64 encoding, of course the raw message body is just a big 
block of ASCII chars encoding the base64 ... notably, with the list 
footer/signature appended below in plain 7-bit ASCII.  I wonder if some 
clients (including pan, it would appear, tho gmane's web interface 
displayed the same body I saw in pan) can't handle the mix of base64 
encoding and plain 7-bit ASCII?, in a message the headers clearly state 
as base64 encoded?  Certainly, the listserv shouldn't just add a plain-
text blob of text to a message body that is base64 encoded like that, 
instead making it multi-part and putting the sig in a 7-bit ascii or 
whatever encoding, if need be, or decoding the base64, appending the text 
blob, and recoding, but regardless, if that's all the "corruption" is, 
clients should handle it better than trucating and/or corrupting.

I didn't try to manually decode the block of base64 either using uudeview 
or the like or fully manually, to see if it stopped at the same spot, or 
included the corruption, or something else, but I do have the full raw 
message file as downloaded via gmane using pan, and can mail it to you 
privately if you'd like.  Tho if you're familiar at all with news and 
have a client to use on news.gmane.org, or simply have lynx around and 
can follow the link above in it, you can find and grab the news article 
for yourself.

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman