wrong values in "df" and "btrfs filesystem df"
Hallo, linux-btrfs,
First I create an array of 2 disks with
mkfs.btrfs -d raid0 -m raid1 /dev/sdb1 /dev/sdd1
and mount it at /srv/MM.
Then I fill it with about 1,6 TByte.
And then I add /dev/sde1 via
btrfs device add /dev/sde1 /srv/MM
btrfs filesystem balance /srv/MM
(it run about 20 hours)
Then I work on it, copy some new files, delete some old files - all works well. Only
df /srv/MM
btrfs filesystem df /srv/MM
show some completely wrong values:
# df /srv/MM
Filesystem 1K-blocks Used Available Use% Mounted on/dev/sdd1 5846053400 1593436456 2898463184 36% /srv/MM
# btrfs filesystem df /srv/MM
Data, RAID0: total=1.67TB, used=1.48TB
System, RAID1: total=16.00MB, used=112.00KB
System: total=4.00MB, used=0.00
Metadata, RAID1: total=3.75GB, used=2.26GB
# btrfs-show
Label: MMedia uuid: 120b036a-883f-46aa-bd9a-cb6a1897c8d2
Total devices 3 FS bytes used 1.48TB devid 3 size 1.81TB used 573.76GB path /dev/sdb1 devid 2 size 1.81TB used 573.77GB path /dev/sde1 devid 1 size 1.82TB used 570.01GB path /dev/sdd1
Btrfs Btrfs v0.19
------------------------------------------------
"df" shows an "Available" value which isn't related to any real value. The sum of "used" and "Available" is far away from the really existent disk space. When I copy additional files to "/srv/MM" then "used" still shows the right value, and the sum grows (slowly) to the max. available space.
In "btrfs filesystem df /srv/MM" the line
Data, RAID0: total=1.67TB, used=1.48TB
shows a "total" value which isn't related to any existent value; maybe it still shows the used space before adding the third partition. This (wrong) value seems not to change.
Kernel 2.6.38.1
btrfs from november 2010
Viele Gruesse!
Helmut
Quoted 43 lines · Helmut Hullen
Hallo, linux-btrfs,
First I create an array of 2 disks with
mkfs.btrfs -d raid0 -m raid1
/dev/sdb1/dev/sdd1and mount it at
/srv/MM.Then I fill it with about 1,6 TByte.
And then I add/dev/sde1viabtrfs device add
/dev/sde1/srv/MM
btrfs filesystem balance/srv/MM
(it run about 20 hours)Then I work on it, copy some new files, delete some old files - all works well. Only
df
/srv/MM
btrfs filesystem df/srv/MMshow some completely wrong values:
# df /srv/MMFilesystem 1K-blocks Used Available Use% Mounted on
/dev/sdd15846053400 1593436456 2898463184 36%/srv/MM# btrfs filesystem df /srv/MMData, RAID0: total=1.67TB, used=1.48TB
System, RAID1: total=16.00MB, used=112.00KB
System: total=4.00MB, used=0.00
Metadata, RAID1: total=3.75GB, used=2.26GB# btrfs-showLabel: MMedia uuid: 120b036a-883f-46aa-bd9a-cb6a1897c8d2
Total devices 3 FS bytes used 1.48TB devid 3 size 1.81TB used 573.76GB path /dev/sdb1 devid 2 size 1.81TB used 573.77GB path /dev/sde1 devid 1 size 1.82TB used 570.01GB path /dev/sdd1
Quoted 5 lines
Btrfs Btrfs v0.19
------------------------------------------------
"df" shows an "Available" value which isn't related to any real value.
I _think_ that value is the amount of space not allocated to any
block group. If that's so, then Available (from df) plus the three "total" values (from btrfs fi df) should equal the size value from df.
Quoted 5 lines
The sum of "used" and "Available" is far away from the really existent disk space. When I copy additional files to "
/srv/MM" then "used" still shows the right value, and the sum grows (slowly) to the max. available space.
Quoted 7 lines
In "btrfs filesystem df
/srv/MM" the lineData, RAID0: total=1.67TB, used=1.48TB
shows a "total" value which isn't related to any existent value; maybe it still shows the used space before adding the third partition. This (wrong) value seems not to change.
It's not wrong -- it simply doesn't mean what you think it does. :)
The total value in the output of "btrfs fi df" is the total space
allocated to block groups. As the filesystem needs more space, it will allocate more block groups from the available raw storage pool, and the number will go up.
This is explained on the wiki at [1].
HTH,
Hugo.
Signature · 4 lines
=== Hugo Mills: hugo@... carfax.org.uk | darksatanic.net | lug.org.uk ===
PGP key: 515C238D from wwwkeys.eu.pgp.net or http://www.carfax.org.uk
--- There are three things you should never see being made: laws, ---
standards, and sausages. 2011-04-09 10:11:41 +0100, Hugo Mills:
[...]
Quoted 22 lines
Quoted 22 lines
# df /srv/MMFilesystem 1K-blocks Used Available Use% Mounted on
/dev/sdd15846053400 1593436456 2898463184 36%/srv/MM# btrfs filesystem df /srv/MMData, RAID0: total=1.67TB, used=1.48TB
System, RAID1: total=16.00MB, used=112.00KB
System: total=4.00MB, used=0.00
Metadata, RAID1: total=3.75GB, used=2.26GB# btrfs-showLabel: MMedia uuid: 120b036a-883f-46aa-bd9a-cb6a1897c8d2
Total devices 3 FS bytes used 1.48TB devid 3 size 1.81TB used 573.76GB path /dev/sdb1 devid 2 size 1.81TB used 573.77GB path /dev/sde1 devid 1 size 1.82TB used 570.01GB path /dev/sdd1Btrfs Btrfs v0.19
Quoted 7 lines
Quoted 3 lines
------------------------------------------------
"df" shows an "Available" value which isn't related to any real value.
I _think_ that value is the amount of space not allocated to any
block group. If that's so, then Available (from df) plus the three "total" values (from btrfs fi df) should equal the size value from df.
[...]
Well,
$ echo $((2898463184 + 1.67*2**30 + 4*2**10 + 16*2**10*2 + 3.75*2**20*2)) 4699513214.0799999
I do get the same kind of discrepancy:
$ df -h /mnt
Filesystem Size Used Avail Use% Mounted on
/dev/sdb 8.2T 3.5T 3.2T 53% /mnt
$ sudo btrfs fi show
Label: none uuid: ...
Total devices 3 FS bytes used 3.43TB
devid 4 size 2.73TB used 1.17TB path /dev/sdc
devid 3 size 2.73TB used 1.17TB path /dev/sdb
devid 2 size 2.70TB used 1.14TB path /dev/sda4
$ sudo btrfs fi df /mnt
Data, RAID0: total=3.41TB, used=3.41TB
System, RAID1: total=16.00MB, used=232.00KB
Metadata, RAID1: total=35.25GB, used=20.55GB$ echo $((3.2 + 3.41 + 2*16/2**20 + 2*35.25/2**10)) 6.6788476562500003
Signature · 1 line
Stephane
Hallo, Hugo,
Du meintest am 09.04.11:
Quoted 29 lines
Quoted 29 lines
df
/srv/MM
btrfs filesystem df/srv/MMshow some completely wrong values:
# df /srv/MMFilesystem 1K-blocks Used Available Use% Mounted on
/dev/sdd15846053400 1593436456 2898463184 36%/srv/MM# btrfs filesystem df /srv/MMData, RAID0: total=3D1.67TB, used=3D1.48TB
System, RAID1: total=3D16.00MB, used=3D112.00KB
System: total=3D4.00MB, used=3D0.00
Metadata, RAID1: total=3D3.75GB, used=3D2.26GB# btrfs-showLabel: MMedia uuid: 120b036a-883f-46aa-bd9a-cb6a1897c8d2
Total devices 3 FS bytes used 1.48TB devid 3 size 1.81TB used 573.76GB path /dev/sdb1 devid 2 size 1.81TB used 573.77GB path /dev/sde1 devid 1 size 1.82TB used 570.01GB path /dev/sdd1Btrfs Btrfs v0.19
------------------------------------------------
Quoted 7 lines
Quoted 2 lines
"df" shows an "Available" value which isn't related to any real value.
I _think_ that value is the amount of space not allocated to any
block group. If that's so, then Available (from df) plus the three "total" values (from btrfs fi df) should equal the size value from df.
I'm not convinced - sorry.
"used" plus "available" should be nearly the same value as the "total" = =20
in "df"; in my example 1.6 TB + 2.9 TB is far away from the total of 5.= 8 =20
TB (here: Tera =3D 10^9).
Quoted 6 lines
The total value in the output of "btrfs fi df" is the total space
allocated to block groups. As the filesystem needs more space, it will allocate more block groups from the available raw storage pool, and the number will go up.This is explained on the wiki at [1].
I've studied the page - the data shown there looks to be consistent. =20
Especially: "used" plus "avail" =3D "size"
-------------------------------------
The data seems (mostly) to be consistent on a new array of disks/=20 partitions. But when I work with "device add" and "device delete" then there are holes.
Another machine:
/dev/sda about 140 GB/dev/sdb about 140 GB/dev/sdc about 70 GB
# mkfs.btrfs -L SCSI -d raid0 -m raid1 /dev/sdb1 /dev/sdc1 # mit 70 GByte gef=FCllt
# df -t btrfs =46ilesystem Type 1K-blocks Used Available Use% Mounted on /dev/sdb1 btrfs 214058944 73956996 65522124 54% /mnt/SCSI
# btrfs filesystem show Label: 'SCSI' uuid: 1932d11d-021a-4054-8429-db25a0204221 Total devices 2 FS bytes used 70.43GB devid 1 size 136.73GB used 37.03GB path /dev/sdb1 devid 2 size 67.41GB used 37.01GB path /dev/sdc1
# btrfs filesystem df /mnt/SCSI Data, RAID0: total=3D72.00GB, used=3D70.33GB Data: total=3D8.00MB, used=3D0.00 System, RAID1: total=3D8.00MB, used=3D12.00KB System: total=3D4.00MB, used=3D0.00 Metadata, RAID1: total=3D1.00GB, used=3D104.78MB Metadata: total=3D8.00MB, used=3D0.00
#------------------------------------------------------------
# btrfs device add /dev/sda1 /mnt/SCSI
# df -t btrfs =46ilesystem Type 1K-blocks Used Available Use% Mounted on /dev/sdb1 btrfs 356234160 84177056 200667344 30% /mnt/SCSI
# btrfs filesystem show Label: 'SCSI' uuid: 1932d11d-021a-4054-8429-db25a0204221 Total devices 3 FS bytes used 80.16GB devid 1 size 136.73GB used 42.03GB path /dev/sdb1 devid 3 size 135.59GB used 0.00 path /dev/sda1 devid 2 size 67.41GB used 42.01GB path /dev/sdc1
# btrfs filesystem df /mnt/SCSI Data, RAID0: total=3D82.00GB, used=3D80.04GB Data: total=3D8.00MB, used=3D0.00 System, RAID1: total=3D8.00MB, used=3D12.00KB System: total=3D4.00MB, used=3D0.00 Metadata, RAID1: total=3D1.00GB, used=3D119.54MB Metadata: total=3D8.00MB, used=3D0.00
#------------------------------------------------------------
# btrfs fi balance /mnt/SCSI
# df -t btrfs =46ilesystem Type 1K-blocks Used Available Use% Mounted on /dev/sdb1 btrfs 356234160 84125344 228211536 27% /mnt/SCSI
# btrfs filesystem show Label: 'SCSI' uuid: 1932d11d-021a-4054-8429-db25a0204221 Total devices 3 FS bytes used 80.14GB devid 1 size 136.73GB used 28.28GB path /dev/sdb1 devid 3 size 135.59GB used 27.00GB path /dev/sda1 devid 2 size 67.41GB used 29.01GB path /dev/sdc1
# btrfs filesystem df /mnt/SCSI Data, RAID0: total=3D81.00GB, used=3D80.04GB Data: total=3D8.00MB, used=3D0.00 System, RAID1: total=3D8.00MB, used=3D12.00KB System: total=3D4.00MB, used=3D0.00 Metadata, RAID1: total=3D1.25GB, used=3D94.29MB Metadata: total=3D8.00MB, used=3D0.00
#------------------------------------------------------------
# btrfs device delete /dev/sdc1 /mnt/SCSI # umount /mnt/SCSI # mount LABEL=3DSCSI /mnt/SCSI
# df -t btrfs =46ilesystem Type 1K-blocks Used Available Use% Mounted on /dev/sdb1 btrfs 285548192 232024076 51758540 82% /mnt/SCSI
# btrfs filesystem show Label: 'SCSI' uuid: 1932d11d-021a-4054-8429-db25a0204221 Total devices 2 FS bytes used 221.04GB devid 1 size 136.73GB used 111.51GB path /dev/sdb1 devid 3 size 135.59GB used 111.51GB path /dev/sda1
# btrfs filesystem df /mnt/SCSI Data, RAID0: total=3D222.00GB, used=3D220.80GB System, RAID1: total=3D8.00MB, used=3D24.00KB System: total=3D4.00MB, used=3D0.00 Metadata, RAID1: total=3D512.00MB, used=3D243.77MB
--------------------------------------------------------------
What "df" shows as "available" immediately after creation is wrong. What it shows after "device add" is wrong.
What it shows after "balance" is wrong.
What it shows after "device delete" looks ok.
Ok - now I presume that sometimes "df" or/and the "btrfs" commands only=
=20
show wrong values, but the system works with the right values (there is= t =20
more free space than shown). But I fear that's not reliable.
--------------------------------
I also write scripts etc; I know since decades it's easy to make =20 mistakes, but sometimes it's hard work to find them.
Viele Gruesse!
Helmut
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" = in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Quoted 24 lines · Hugo Mills
Quoted 15 lines · Helmut Hullen
Hallo, linux-btrfs,
First I create an array of 2 disks with
mkfs.btrfs -d raid0 -m raid1
/dev/sdb1/dev/sdd1and mount it at
/srv/MM.Then I fill it with about 1,6 TByte.
And then I add/dev/sde1viabtrfs device add
/dev/sde1/srv/MM
btrfs filesystem balance/srv/MM
(it run about 20 hours)Quoted 7 lines
Then I work on it, copy some new files, delete some old files - all works well. Only
df
/srv/MM
btrfs filesystem df/srv/MMshow some completely wrong values:
Quoted 8 lines
It's not wrong -- it simply doesn't mean what you think it does. :)
The total value in the output of "btrfs fi df" is the total space
allocated to block groups. As the filesystem needs more space, it will allocate more block groups from the available raw storage pool, and the number will go up.This is explained on the wiki at [1].
And I just drew up a picture which I think should help explain it a bit,
too: http://www.kepstin.ca/dump/btrfs-alloc.png
If I can figure out how to add images to the btrfs wiki, and find a good place to put it, do you think this would be a helpful addition?
Quoted 1 line
Signature · 1 line
Calvin Walton <[email protected]>
Hallo, Calvin,
Du meintest am 09.04.11:
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Then I work on it, copy some new files, delete some old files - all works well. Only
df
/srv/MM
btrfs filesystem df/srv/MMshow some completely wrong values:
[...]
Quoted 2 lines
And I just drew up a picture which I think should help explain it a bit, too: http://www.kepstin.ca/dump/btrfs-alloc.png
Nice picture. But it doesn't solve the problem that I need a reliable information about the free/available space. And I prefer asking with "df" for this information - "df" should work in the same way for all filesystems.
Viele Gruesse!
Helmut
Quoted 8 lines · Helmut Hullen
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Then I work on it, copy some new files, delete some old files - all works well. Only
df
/srv/MM
btrfs filesystem df/srv/MMshow some completely wrong values:
Quoted 7 lines
Quoted 2 lines
And I just drew up a picture which I think should help explain it a bit, too: http://www.kepstin.ca/dump/btrfs-alloc.png
Nice picture. But it doesn't solve the problem that I need a reliable information about the free/available space. And I prefer asking with "df" for this information - "df" should work in the same way for all filesystems.
The problem is that the answer to the seemingly simple question: "How much more data can I put onto this filesystem?" gets pretty hard with btrfs.
Your case is one of the simpler ones - To calculate the remaining space for files, you take the unused allocated data space (light blue on my picture), add the unallocated space (white), divide by the raid mode redundancy, and subtract some percentage (this is only an estimate, of course...) of that unallocated space for the additional metadata overhead.
Now imagine the case where your btrfs filesystem has files stored in multiple raid modes: e.g. some files are raid5, others raid0. The amount of data you can write to the filesystem then depends on how you write the data!
You might be able to fit 64gb if you use raid0, but only 48gb with raid5; and only 16gb with raid1!
There isn't a single number that btrfs can report which does what you want.
Signature · 1 line
Calvin Walton <[email protected]>
Hallo, Calvin,
Du meintest am 09.04.11:
Quoted 9 lines
Quoted 4 lines
Nice picture. But it doesn't solve the problem that I need a reliable information about the free/available space. And I prefer asking with "df" for this information - "df" should work in the same way for all filesystems.
The problem is that the answer to the seemingly simple question: "How much more data can I put onto this filesystem?" gets pretty hard with btrfs.
Quoted 6 lines
Your case is one of the simpler ones - To calculate the remaining space for files, you take the unused allocated data space (light blue on my picture), add the unallocated space (white), divide by the raid mode redundancy, and subtract some percentage (this is only an estimate, of course...) of that unallocated space for the additional metadata overhead.
That's simple?
Maybe I'm simple minded. But I expect the same meaning of the shown date with
df <mointpoint>
regardless what kind of filesystem is mounted.
But no test item for an IQ test.
If the value of "available" is unresolvable then btrfs should not show any value.
Ok - there's a problem slightly similar to the "available" value of compressed partitions.
Viele Gruesse!
Helmut
Quoted 2 lines · Helmut Hullen
If the value of "available" is unresolvable then btrfs should not show any value.
Disagree strongly. I think a pessimistic estimate would be much better to show, than no value at all. This may be what is currently shown.
As for solving this with a high degree of usability, that's not really possible when constrained to the traditional paradigm that one filesystem will have completely consistent behavior backing all of it.
I think the only answer is to have btrfs-specific tools that know more about the filesystem, and can present the relevant facts.
Taking the example of part fs being raid0 and part being raid5, such a tool would then list calculated values for both those parts of the fs. One showing how much could go into the raid0 part, the other how much could go into the raid5 part.
But for such filesystems, Linux can't do what Helmut would like.
Maybe it would be possible to optimize the reported numbers, to be what the user actually wants as often as possible. Ie. if there is only one type of backing storage (sorry, don't know the terms) then the calculation would be easier to get right, following the simple formula that was just given. This is all eye candy however, completely irrelevant IMO as long as the filesystem oopses, or eats root nodes. :)
//Peter
Quoted 1 line · Peter Stuge
But for such filesystems, Linux can't do what Helmut would like.
Indeed. In the general case, there's no single number that "gets it
right". That said, if data and metadata are the same replication type for the whole filesystem, you *can* get it pretty much exactly right(*); if data is the same type for the whole filesystem, you can get it *nearly* right (since metadata is typically a small fraction of the data size).
(*) Ignoring the space required for reservations to actually do anything with the filesystem when it's very close to full...
Quoted 3 lines
Maybe it would be possible to optimize the reported numbers, to be what the user actually wants as often as possible. Ie. if there is only one type of backing storage (sorry, don't know the terms) then
RAID level or replication strategy. Although "RAID" gives a
slightly misleading impression of what actually happens in btrfs...
Quoted 7 lines
the calculation would be easier to get right, following the simple formula that was just given. This is all eye candy however, completely irrelevant IMO as long as the filesystem oopses, or eats root nodes. :)
//Peter
Signature · 3 lines
=== Hugo Mills: hugo@... carfax.org.uk | darksatanic.net | lug.org.uk === PGP key: 515C238D from wwwkeys.eu.pgp.net or http://www.carfax.org.uk --- Great oxymorons of the world, no. 3: Military Intelligence ---
Quoted 46 lines · Hugo Mills
Quoted 45 lines · Helmut Hullen
Hallo, linux-btrfs,
First I create an array of 2 disks with
mkfs.btrfs -d raid0 -m raid1
/dev/sdb1/dev/sdd1and mount it at
/srv/MM.Then I fill it with about 1,6 TByte.
And then I add/dev/sde1viabtrfs device add
/dev/sde1/srv/MM
btrfs filesystem balance/srv/MM
(it run about 20 hours)Then I work on it, copy some new files, delete some old files - all works well. Only
df
/srv/MM
btrfs filesystem df/srv/MMshow some completely wrong values:
# df /srv/MMFilesystem 1K-blocks Used Available Use% Mounted on
/dev/sdd15846053400 1593436456 2898463184 36%/srv/MM# btrfs filesystem df /srv/MMData, RAID0: total=1.67TB, used=1.48TB
System, RAID1: total=16.00MB, used=112.00KB
System: total=4.00MB, used=0.00
Metadata, RAID1: total=3.75GB, used=2.26GB# btrfs-showLabel: MMedia uuid: 120b036a-883f-46aa-bd9a-cb6a1897c8d2
Total devices 3 FS bytes used 1.48TB devid 3 size 1.81TB used 573.76GB path /dev/sdb1 devid 2 size 1.81TB used 573.77GB path /dev/sde1 devid 1 size 1.82TB used 570.01GB path /dev/sdd1Btrfs Btrfs v0.19
Quoted 7 lines
Quoted 3 lines
------------------------------------------------
"df" shows an "Available" value which isn't related to any real value.
I _think_ that value is the amount of space not allocated to any
block group. If that's so, then Available (from df) plus the three "total" values (from btrfs fi df) should equal the size value from df.
This value excludes the space that can not be allocated to any block group, This feature was implemented to fix the bug df command add the disk space, which can not be allocated to any block group forever, into the "Available" value. (see the changelog of the commit 6d07bcec969af335d4e35b3921131b7929bd634e)
This implementation just like fake chunk allocation, but the fake allocation just allocate the space from two of these three disks, doesn't spread the stripes over all the disks, which has enough space.
I will post a patch to fix it soon.
Quoted 28 lines
Quoted 6 lines
The sum of "used" and "Available" is far away from the really existent disk space. When I copy additional files to "
/srv/MM" then "used" still shows the right value, and the sum grows (slowly) to the max. available space.Quoted 7 lines
In "btrfs filesystem df
/srv/MM" the lineData, RAID0: total=1.67TB, used=1.48TB
shows a "total" value which isn't related to any existent value; maybe it still shows the used space before adding the third partition. This (wrong) value seems not to change.
It's not wrong -- it simply doesn't mean what you think it does. :)
The total value in the output of "btrfs fi df" is the total space
allocated to block groups. As the filesystem needs more space, it will allocate more block groups from the available raw storage pool, and the number will go up.This is explained on the wiki at [1].
HTH,
Hugo.
2011-04-10 18:13:51 +0800, Miao Xie:
[...]
Quoted 30 lines
Quoted 22 lines
Quoted 22 lines
# df /srv/MMFilesystem 1K-blocks Used Available Use% Mounted on
/dev/sdd15846053400 1593436456 2898463184 36%/srv/MM# btrfs filesystem df /srv/MMData, RAID0: total=1.67TB, used=1.48TB
System, RAID1: total=16.00MB, used=112.00KB
System: total=4.00MB, used=0.00
Metadata, RAID1: total=3.75GB, used=2.26GB# btrfs-showLabel: MMedia uuid: 120b036a-883f-46aa-bd9a-cb6a1897c8d2
Total devices 3 FS bytes used 1.48TB devid 3 size 1.81TB used 573.76GB path /dev/sdb1 devid 2 size 1.81TB used 573.77GB path /dev/sde1 devid 1 size 1.82TB used 570.01GB path /dev/sdd1Btrfs Btrfs v0.19
Quoted 7 lines
Quoted 3 lines
------------------------------------------------
"df" shows an "Available" value which isn't related to any real value.
I _think_ that value is the amount of space not allocated to any
block group. If that's so, then Available (from df) plus the three "total" values (from btrfs fi df) should equal the size value from df.
Quoted 8 lines
This value excludes the space that can not be allocated to any block group, This feature was implemented to fix the bug df command add the disk space, which can not be allocated to any block group forever, into the "Available" value. (see the changelog of the commit
6d07bcec969af335d4e35b3921131b7929bd634e)This implementation just like fake chunk allocation, but the fake allocation just allocate the space from two of these three disks, doesn't spread the stripes over all the disks, which has enough space.
[...]
Hi Miao,
would you care to expand a bit on that. In Helmut's case above where all the drives have at least 1.2TB free, how would there be un-allocatable space?
What's the implication of having disks of differing sizes? Does that mean that the extra space on larger disks is lost?
Thanks,
Stephane
Quoted 6 lines · Stephane Chazelas
2011-04-10 18:13:51 +0800, Miao Xie:[...]
What's the implication of having disks of differing sizes? Does that mean that the extra space on larger disks is lost?
Yes. Currently the allocator cannot handle different sizes well, especially when mirroring is involved. I sent a patch for this to the list some weeks ago (see quasi-round-robin), but it hasn't been merged yet.
-Arne
Quoted 3 lines
Thanks,
Stephane
Hallo, Stephane,
Du meintest am 11.04.11:
Quoted 2 lines
What's the implication of having disks of differing sizes? Does that mean that the extra space on larger disks is lost?
Seems to work.
I've tried:
/dev/sda 140 GByte/dev/sdb 140 GByte/dev/sdc 70 GByte
mkfs.btrfs -d raid0 -m raid1 /dev/sdb1 /dev/sdc1
mounted, more than 140 GByte free
Filled with more than 140 GByte (/dev/sdc1 was full to the brim)
btrfs device add /dev/sda1 ...
btrfs filesystem balance ...
Needed many hours, but then more than 210 GByte were usable.
Filled up to about 220 GByte; /dev/sdc1 was again full to the brim
btrfs device delete /dev/sdc1
umount
mount
All looks as expected, only the 2 bigger devices are seen, and they contain the expected files.
And that looks good: my major interest in btrfs is working in that way - adding a bigger device, deleting a smaller device.
Kernel 2.6.38.1
btrfs from november 2010
Only the values shown with "df" and "btrfs filesystem df" need getting used to; maybe "available" has to be seen as "at least available".
Viele Gruesse!
Helmut
Quoted 42 lines · Stephane Chazelas
2011-04-10 18:13:51 +0800, Miao Xie:[...]
Quoted 38 lines
Quoted 22 lines
Quoted 22 lines
# df /srv/MMFilesystem 1K-blocks Used Available Use% Mounted on
/dev/sdd15846053400 1593436456 2898463184 36%/srv/MM# btrfs filesystem df /srv/MMData, RAID0: total=1.67TB, used=1.48TB
System, RAID1: total=16.00MB, used=112.00KB
System: total=4.00MB, used=0.00
Metadata, RAID1: total=3.75GB, used=2.26GB# btrfs-showLabel: MMedia uuid: 120b036a-883f-46aa-bd9a-cb6a1897c8d2
Total devices 3 FS bytes used 1.48TB devid 3 size 1.81TB used 573.76GB path /dev/sdb1 devid 2 size 1.81TB used 573.77GB path /dev/sde1 devid 1 size 1.82TB used 570.01GB path /dev/sdd1Btrfs Btrfs v0.19
Quoted 7 lines
Quoted 3 lines
------------------------------------------------
"df" shows an "Available" value which isn't related to any real value.
I _think_ that value is the amount of space not allocated to any
block group. If that's so, then Available (from df) plus the three "total" values (from btrfs fi df) should equal the size value from df.This value excludes the space that can not be allocated to any block group, This feature was implemented to fix the bug df command add the disk space, which can not be allocated to any block group forever, into the "Available" value. (see the changelog of the commit
6d07bcec969af335d4e35b3921131b7929bd634e)This implementation just like fake chunk allocation, but the fake allocation just allocate the space from two of these three disks, doesn't spread the stripes over all the disks, which has enough space.
[...]
Quoted 8 lines
Hi Miao,
would you care to expand a bit on that. In Helmut's case above where all the drives have at least 1.2TB free, how would there be un-allocatable space?
What's the implication of having disks of differing sizes? Does that mean that the extra space on larger disks is lost?
I'm sorry that I couldn't explain it clearly.
As we know, Btrfs introduced RAID fucntion, and it can allocate some stripes from different disks to make up a RAID block group. But if there is not enough disk space to allocate enough stripes, btrfs can't make up a new block group, and the left disk space can't be used forever.
For example, If we have two disks, one is 5GB, and the other is 10GB, and we use RAID0 block groups to store the file data. The RAID0 block group needs two stripes which are on the different disks at least. After all space on the 5GB disk is allocated, there is about 5GB free space on the 10GB disk, this space can not be used because we have no free space on the other disk to allocate, and can't make up a new RAID0 block group.
Beside the two-stripe limit, the chunk allocator will allocate stripes from every disk as much as possible, to make up a new RAID0 block group. That is if all the disks have enough free space, the allocator will allocate stripes from all the disks.
In Helmut's case, the chunk allocator will allocate three same-size stripes from those three disks to make up the new RAID0 block group, every time btrfs allocate new chunks (block groups), until there is no free space on two disks. So btrfs can use most of the disk space for RAID0 block group.
But the algorithm of df command doesn't simulate the above allocation correctly, this simulated allocation just allocates the stripes from two disks, and then, these two disks have no free space, but the third disk still has 1.2TB free space, df command thinks this space can be used to make a new RAID0 block group and ignores it. This is a bug, I think.
BTW: "Available" value is the size of the free space that we may use it to store the file data. In btrfs filesystem, it is hard to calculate, because the block groups are allocated dynamically, not all the free space on the disks is allocated to make up data block groups, some of the space may be allocated to make up data block groups. So we just tell the users the size of free space maybe they can use to store the file data.
Thanks
Miao
Quoted 4 lines
Thanks,
Stephane
2011-04-12 15:22:57 +0800, Miao Xie:
[...]
Quoted 5 lines
But the algorithm of df command doesn't simulate the above allocation correctly, this simulated allocation just allocates the stripes from two disks, and then, these two disks have no free space, but the third disk still has 1.2TB free space, df command thinks this space can be used to make a new RAID0 block group and ignores it. This is a bug, I think.
[...]
Thanks a lot Miao for the detailed explanation. So, the disk space is not lost, it's just df not reporting the available space correctly. That's me relieved.
It explains why I'm getting:
# blockdev --getsize64 /dev/sda4 2967698087424 # blockdev --getsize64 /dev/sdb 3000592982016 # blockdev --getsize64 /dev/sdc 3000592982016 # truncate -s 2967698087424 a # truncate -s 3000592982016 b # truncate -s 3000592982016 c # losetup /dev/loop0 ./a # losetup /dev/loop1 ./b # losetup /dev/loop2 ./c # mkfs.btrfs a b c # btrfs device scan /dev/loop[0-2] Scanning for Btrfs filesystems in '/dev/loop0' Scanning for Btrfs filesystems in '/dev/loop1' Scanning for Btrfs filesystems in '/dev/loop2' # mount /dev/loop0 /mnt/1 # df -k /mnt/1 Filesystem 1K-blocks Used Available Use% Mounted on /dev/loop0 8758675828 56 5859474304 1% /mnt/1 # echo $(((8758675828 - 5859474304)*2**10)) 2968782360576
One disk worth of space lost according to df.
While it should have been more something like $(((3000592982016-2967698087424)*2)) (about 60GB), or about 0 after the quasi-round-robin allocation patch, right?
Best regards,
Stephane
Quoted 39 lines · Stephane Chazelas
2011-04-12 15:22:57 +0800, Miao Xie:[...]
Quoted 5 lines
But the algorithm of df command doesn't simulate the above allocation correctly, this simulated allocation just allocates the stripes from two disks, and then, these two disks have no free space, but the third disk still has 1.2TB free space, df command thinks this space can be used to make a new RAID0 block group and ignores it. This is a bug, I think.
[...]
Thanks a lot Miao for the detailed explanation. So, the disk space is not lost, it's just df not reporting the available space correctly. That's me relieved.
It explains why I'm getting:
# blockdev --getsize64 /dev/sda4 2967698087424 # blockdev --getsize64 /dev/sdb 3000592982016 # blockdev --getsize64 /dev/sdc 3000592982016 # truncate -s 2967698087424 a # truncate -s 3000592982016 b # truncate -s 3000592982016 c # losetup /dev/loop0 ./a # losetup /dev/loop1 ./b # losetup /dev/loop2 ./c # mkfs.btrfs a b c # btrfs device scan /dev/loop[0-2] Scanning for Btrfs filesystems in '/dev/loop0' Scanning for Btrfs filesystems in '/dev/loop1' Scanning for Btrfs filesystems in '/dev/loop2' # mount /dev/loop0 /mnt/1 # df -k /mnt/1 Filesystem 1K-blocks Used Available Use% Mounted on /dev/loop0 8758675828 56 5859474304 1% /mnt/1 # echo $(((8758675828 - 5859474304)*2**10)) 2968782360576
Quoted 5 lines
One disk worth of space lost according to df.
While it should have been more something like $(((3000592982016-2967698087424)*2)) (about 60GB), or about 0 after the quasi-round-robin allocation patch, right?
I think btrfs can utilize the disk space very well(perhaps all the space). The last 60GB is on the different disks, so btrfs can utilize the space to make up new RAID0 block groups with two stripes.
Quoted 7 lines
Best regards,
Stephane
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html