Re: ENOSPC during balance

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 




i did a `btrfs file convert -dconvert=raid10,soft /data`, which converted the whole filesystem back to raid10, it completed without errors.
then i did a `btrfs convert -dconvert=raid1 /data`, which completed with 184 ENOSPC errors, exactly the same amount as before the covert back to raid10



On 13 Jan 2014, at 18:43, David Sterba <dsterba@xxxxxxx> wrote:

> On Sun, Jan 12, 2014 at 03:49:12PM +0100, Remco Hosman - Yerf IT wrote:
>> I am trying to convert my array from raid10 to 1, and its partially
>> completed, but at the moment i am getting a '59366.459092] btrfs: 185
>> enospc errors during balance’ when i try to balance anything more with
>> `btrfs bal start -dconvert=raid1,soft /mountpoint`
>> 
>> I have already scanned for files with extends over 1gig gig, and there
>> is at least 100gig unallocated on each of the disks, and scrub reports
>> no error at all.
>> Kernel is 3.13-rc7 and tools are latest from git.
> 
> By unalocated you mean from 'btrfs fi df' output "total - used = 100G"
> or that the sum of all occupied space is 100G less than the device size?
> So if there's some space left to allocate new 1G-chunks for balance.
> 
> How many disks does the fs contain?
> 

filesystem is 6 disks of varying sizes.

>> Anything else i can try ?
> 
> Run
> $ btrfs balance start -dusage=0,profiles=raid10\|raid1 /mnt
> 
> if there are some chunks preallocated from previous balance runs, this
> will clean them. The -musage=0 filter could also get some space.
> 
> I've experienced similar problems with conversion from raid1 to raid10,
> where it's probably worse regarding the 1G-chunks, because the raid-0
> level needs the chunk on each disk, while raid1 is fine with just 2.
> 
> I had done the -dusage=0 cleanup step every time the 'enospc during
> balance' was hit, and it finished in the end. Not perfect, an automatic
> and more intelligent chunk reclaim is among the project ideas though.
> 
> If nothing from above helps, please post the output of 'fi df' and 'fi
> show' commands.
> 

i did a `btrfs bal start -dconvert=raid10,soft /data`, and it completed without errors.
then a `btrfs bal start -donvert=raid1 /data`, which resulted in 184 ENOSPC errors.

currently, the filesystem looks like this:

Data, RAID10: total=431.57GiB, used=430.22GiB
Data, RAID1: total=5.18TiB, used=5.18TiB
System, RAID10: total=96.00MiB, used=804.00KiB
Metadata, RAID10: total=12.38GiB, used=9.34GiB

Label: data  uuid: a8626d67-4684-4b23-99b3-8d5fa8e7fd69
	Total devices 6 FS bytes used 5.61TiB
	devid    1 size 1.82TiB used 1.27TiB path /dev/sdg2
	devid    2 size 1.82TiB used 1.27TiB path /dev/sdb2
	devid    3 size 1.82TiB used 1.27TiB path /dev/sdf2
	devid    5 size 2.73TiB used 2.17TiB path /dev/sdd2
	devid    10 size 2.73TiB used 2.18TiB path /dev/sde2
	devid    11 size 3.64TiB used 3.08TiB path /dev/sdc1

when i do a `btrfs bal start -dconvert=raid1,soft /data` :
dmesg: [560325.834835] btrfs: 184 enospc errors during balance

Data, RAID10: total=428.57GiB, used=428.57GiB
Data, RAID1: total=5.72TiB, used=5.18TiB
System, RAID10: total=96.00MiB, used=880.00KiB
Metadata, RAID10: total=12.38GiB, used=9.34GiB

Label: data  uuid: a8626d67-4684-4b23-99b3-8d5fa8e7fd69
	Total devices 6 FS bytes used 5.61TiB
	devid    1 size 1.82TiB used 1.45TiB path /dev/sdg2
	devid    2 size 1.82TiB used 1.44TiB path /dev/sdb2
	devid    3 size 1.82TiB used 1.44TiB path /dev/sdf2
	devid    5 size 2.73TiB used 2.35TiB path /dev/sdd2
	devid    10 size 2.73TiB used 2.36TiB path /dev/sde2
	devid    11 size 3.64TiB used 3.26TiB path /dev/sdc1

so it looks like it did allocate all the space it needed, but still failed

kernel is currently 3.13-rc7

Remco


> 
> david



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




[Index of Archives]     [Linux Filesystem Development]     [Linux NFS]     [Linux NILFS]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux