Status: converting raid levels

Status: converting raid levels

[Marcel Ritter]

Hi,

please, can someone comment on the current status of raid level migration?
(kernel 4.0.0-rc2, btrfs-progs 3.19-rc2)

I just started testing this feature, and it doesn't seem to work:

Starting with Raid1:

root@thunder[ ~/btrfs-progs ]# ./btrfs fi df /tmp/m
Data, RAID1: total=3.00GiB, used=512.00KiB
Data, single: total=8.00MiB, used=0.00B
System, RAID1: total=8.00MiB, used=16.00KiB
System, single: total=4.00MiB, used=0.00B
Metadata, RAID1: total=1.00GiB, used=112.00KiB
Metadata, single: total=8.00MiB, used=0.00B
GlobalReserve, single: total=16.00MiB, used=0.00B

Converting to Raid10:

root@thunder[ ~/btrfs-progs ]# ./btrfs balance start -dconvert=raid10
-mconvert=raid10 /tmp/m/
Done, had to relocate 9 out of 9 chunks

Still Raid1 ...:

root@thunder[ ~/btrfs-progs ]# ./btrfs fi df /tmp/m
Data, RAID1: total=35.00GiB, used=31.49GiB
System, RAID1: total=32.00MiB, used=16.00KiB
Metadata, RAID1: total=1.00GiB, used=33.86MiB
GlobalReserve, single: total=16.00MiB, used=0.00B

I also tried conversion of raid6 - but it did not work either
(according to btrfs fi df output).

Did I miss something, or is it a bug?

Bye,
    Marcel

Re: Status: converting raid levels

[Holger Hoffstätte]

On Fri, 06 Mar 2015 11:22:09 +0100, Marcel Ritter wrote:

> please, can someone comment on the current status of raid level migration?
> (kernel 4.0.0-rc2, btrfs-progs 3.19-rc2)
> 
> I just started testing this feature, and it doesn't seem to work:

You're very likely looking at something I found previously, see:
http://www.spinics.net/lists/linux-btrfs/msg42084.html & followups.

So far I haven't seen any fix appear anywhere; if you're daring
enough you can try to revert/undo the mentioned patch and see
what happens when you try to convert between RAID levels.

-h

Re: Status: converting raid levels

[Filipe David Manana]

On Fri, Mar 6, 2015 at 10:22 AM, Marcel Ritter <ritter.marcel@gmail.com> wrote:
> Hi,
>
> please, can someone comment on the current status of raid level migration?
> (kernel 4.0.0-rc2, btrfs-progs 3.19-rc2)
>
> I just started testing this feature, and it doesn't seem to work:
>
> Starting with Raid1:
>
> root@thunder[ ~/btrfs-progs ]# ./btrfs fi df /tmp/m
> Data, RAID1: total=3.00GiB, used=512.00KiB
> Data, single: total=8.00MiB, used=0.00B
> System, RAID1: total=8.00MiB, used=16.00KiB
> System, single: total=4.00MiB, used=0.00B
> Metadata, RAID1: total=1.00GiB, used=112.00KiB
> Metadata, single: total=8.00MiB, used=0.00B
> GlobalReserve, single: total=16.00MiB, used=0.00B
>
> Converting to Raid10:
>
> root@thunder[ ~/btrfs-progs ]# ./btrfs balance start -dconvert=raid10
> -mconvert=raid10 /tmp/m/
> Done, had to relocate 9 out of 9 chunks
>
> Still Raid1 ...:
>
> root@thunder[ ~/btrfs-progs ]# ./btrfs fi df /tmp/m
> Data, RAID1: total=35.00GiB, used=31.49GiB
> System, RAID1: total=32.00MiB, used=16.00KiB
> Metadata, RAID1: total=1.00GiB, used=33.86MiB
> GlobalReserve, single: total=16.00MiB, used=0.00B
>
> I also tried conversion of raid6 - but it did not work either
> (according to btrfs fi df output).
>
> Did I miss something, or is it a bug?

Seems like the regression Holger found due to a patch added to 4.0:

http://www.spinics.net/lists/linux-btrfs/msg42084.html

(he mentions a 3.18.x kernel, but that's because he builds his own
kernels with patches backported)

>
> Bye,
>     Marcel
> --
> To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html



-- 
Filipe David Manana,

"Reasonable men adapt themselves to the world.
 Unreasonable men adapt the world to themselves.
 That's why all progress depends on unreasonable men."

Re: Status: converting raid levels

[Marcel Ritter]

Hi,

I tried to revert the mentioned patch set (kernel 4.0.0-rc2).

Starting a new re-balance with this kernel, while running my I/O test
(big dd write) on the same btrfs volume (14 disks) resulted in "cpu stuck"
messages - system was unusable just a few seconds later.

With a plain 4.0.0-rc2 kernel the rebalance at least does finish
without crashing/hanging the system, but - on the downside - it
does not change the raid layout of the btrfs volume :-(

Any other ideas?

2015-03-07 10:39 GMT+01:00 Filipe David Manana <fdmanana@gmail.com>:
> On Fri, Mar 6, 2015 at 10:22 AM, Marcel Ritter <ritter.marcel@gmail.com> wrote:
>> Hi,
>>
>> please, can someone comment on the current status of raid level migration?
>> (kernel 4.0.0-rc2, btrfs-progs 3.19-rc2)
>>
>> I just started testing this feature, and it doesn't seem to work:
>>
>> Starting with Raid1:
>>
>> root@thunder[ ~/btrfs-progs ]# ./btrfs fi df /tmp/m
>> Data, RAID1: total=3.00GiB, used=512.00KiB
>> Data, single: total=8.00MiB, used=0.00B
>> System, RAID1: total=8.00MiB, used=16.00KiB
>> System, single: total=4.00MiB, used=0.00B
>> Metadata, RAID1: total=1.00GiB, used=112.00KiB
>> Metadata, single: total=8.00MiB, used=0.00B
>> GlobalReserve, single: total=16.00MiB, used=0.00B
>>
>> Converting to Raid10:
>>
>> root@thunder[ ~/btrfs-progs ]# ./btrfs balance start -dconvert=raid10
>> -mconvert=raid10 /tmp/m/
>> Done, had to relocate 9 out of 9 chunks
>>
>> Still Raid1 ...:
>>
>> root@thunder[ ~/btrfs-progs ]# ./btrfs fi df /tmp/m
>> Data, RAID1: total=35.00GiB, used=31.49GiB
>> System, RAID1: total=32.00MiB, used=16.00KiB
>> Metadata, RAID1: total=1.00GiB, used=33.86MiB
>> GlobalReserve, single: total=16.00MiB, used=0.00B
>>
>> I also tried conversion of raid6 - but it did not work either
>> (according to btrfs fi df output).
>>
>> Did I miss something, or is it a bug?
>
> Seems like the regression Holger found due to a patch added to 4.0:
>
> http://www.spinics.net/lists/linux-btrfs/msg42084.html
>
> (he mentions a 3.18.x kernel, but that's because he builds his own
> kernels with patches backported)
>
>>
>> Bye,
>>     Marcel
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
>
>
> --
> Filipe David Manana,
>
> "Reasonable men adapt themselves to the world.
>  Unreasonable men adapt the world to themselves.
>  That's why all progress depends on unreasonable men."