wrong / too less space on raid

wrong / too less space on raid

[source]

hello buglist,

before sending this mail, i tried to track down previous messages from
the archive and found no similar entry. so either something went wrong
on my machine only or maybe there are too less testers around.

about the real problem/bug :

i tried to create several btrfs filesystems with success and tests about
performance and reliability but the free amount given to the system is
far too small, one example :

#mkfs.btrfs -m raid0 -d raid0 /dev/sda8 /dev/sdb8 /dev/sdc8

WARNING! - Btrfs Btrfs v0.19 IS EXPERIMENTAL
WARNING! - see http://btrfs.wiki.kernel.org before using

adding device /dev/sdb8 id 2
adding device /dev/sdc8 id 3
fs created label (null) on /dev/sda8
	nodesize 4096 leafsize 4096 sectorsize 4096 size 131.44GB
Btrfs Btrfs v0.19

# mkdir /test && mount /dev/sda8 /test
# df
Filesystem     1K-blocks      Used Available Use% Mounted on
/dev/md0       292911488 174222808 118688680  60% /
tmpfs               5120         4      5116   1% /lib/init/rw
tmpfs             611976       544    611432   1% /run
udev               10240         0     10240   0% /dev
tmpfs            1223952        76   1223876   1% /run/shm
/dev/md2       292833568     33312 292800256   1% /or
/dev/md1       292833568 202533276  90300292  70% /root/workplace
/dev/sdc1          90297     52782     32698  62% /boot
myon400:/      976694528 485310720 491383808  50% /srv
/dev/sda8      137821156        28  90827392   1% /test

in this case, you can see a raid0 setup for meta and data, also
acknowledged by "sdb8 id2" and "sdc8 id3", the overall amount
of 1k blocks in df = 137821156 but the available amount is 
only 90827392 1-k blocks while 28 are in use. this looks like
the calculation of raid0 outputs something i would assume 
for raid5 or in other words just 2 of 3 raid-members usable.

maybe i misinterpreted something or made a mistake by creation
however any documentation around gave me no idea.

this test was made on a debian sid, using btrfs v0.19.

Re: wrong / too less space on raid

[Miao Xie]

On sun, 27 Nov 2011 13:45:22 +0100, source wrote:
> before sending this mail, i tried to track down previous messages from
> the archive and found no similar entry. so either something went wrong
> on my machine only or maybe there are too less testers around.
> 
> about the real problem/bug :
> 
> i tried to create several btrfs filesystems with success and tests about
> performance and reliability but the free amount given to the system is
> far too small, one example :
> 
> #mkfs.btrfs -m raid0 -d raid0 /dev/sda8 /dev/sdb8 /dev/sdc8
> 
> WARNING! - Btrfs Btrfs v0.19 IS EXPERIMENTAL
> WARNING! - see http://btrfs.wiki.kernel.org before using
> 
> adding device /dev/sdb8 id 2
> adding device /dev/sdc8 id 3
> fs created label (null) on /dev/sda8
> 	nodesize 4096 leafsize 4096 sectorsize 4096 size 131.44GB
> Btrfs Btrfs v0.19
> 
> # mkdir /test && mount /dev/sda8 /test
> # df
> Filesystem     1K-blocks      Used Available Use% Mounted on
> /dev/md0       292911488 174222808 118688680  60% /
> tmpfs               5120         4      5116   1% /lib/init/rw
> tmpfs             611976       544    611432   1% /run
> udev               10240         0     10240   0% /dev
> tmpfs            1223952        76   1223876   1% /run/shm
> /dev/md2       292833568     33312 292800256   1% /or
> /dev/md1       292833568 202533276  90300292  70% /root/workplace
> /dev/sdc1          90297     52782     32698  62% /boot
> myon400:/      976694528 485310720 491383808  50% /srv
> /dev/sda8      137821156        28  90827392   1% /test
> 
> in this case, you can see a raid0 setup for meta and data, also
> acknowledged by "sdb8 id2" and "sdc8 id3", the overall amount
> of 1k blocks in df = 137821156 but the available amount is 
> only 90827392 1-k blocks while 28 are in use. this looks like
> the calculation of raid0 outputs something i would assume 
> for raid5 or in other words just 2 of 3 raid-members usable.
> 
> maybe i misinterpreted something or made a mistake by creation
> however any documentation around gave me no idea.

This number is just a appraised number, not rigorous. It tells us
how much space we can use to make up raid0 block groups which are
used to store the file data.

Thanks
Miao

> 
> this test was made on a debian sid, using btrfs v0.19.
> 
> --
> 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
> 

Re: wrong / too less space on raid

[Phillip Susi]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 11/27/2011 08:36 PM, Miao Xie wrote:
> This number is just a appraised number, not rigorous. It tells us
> how much space we can use to make up raid0 block groups which are
> used to store the file data.

More specifically, the available space that df reports is the amount of space available to store file data, which means it doesn't count the space reserved for metadata block groups, even though a not insignificant amount of that space may be free for metadata allocations.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAk7TB4AACgkQJ4UciIs+XuLdOACaA3hrErj82plyPOxXEoM+kHd3
5XkAnR3vR+0bhGlewjKTSEsVNZgc0mka
=UNVc
-----END PGP SIGNATURE-----

Re: wrong / too less space on raid

[Miao Xie]

On sun, 27 Nov 2011 13:45:22 +0100, source wrote:
> i tried to create several btrfs filesystems with success and tests about
> performance and reliability but the free amount given to the system is
> far too small, one example :
> 
> #mkfs.btrfs -m raid0 -d raid0 /dev/sda8 /dev/sdb8 /dev/sdc8
> 
> WARNING! - Btrfs Btrfs v0.19 IS EXPERIMENTAL
> WARNING! - see http://btrfs.wiki.kernel.org before using
> 
> adding device /dev/sdb8 id 2
> adding device /dev/sdc8 id 3
> fs created label (null) on /dev/sda8
> 	nodesize 4096 leafsize 4096 sectorsize 4096 size 131.44GB
> Btrfs Btrfs v0.19
> 
> # mkdir /test && mount /dev/sda8 /test
> # df
> Filesystem     1K-blocks      Used Available Use% Mounted on
> /dev/md0       292911488 174222808 118688680  60% /
> tmpfs               5120         4      5116   1% /lib/init/rw
> tmpfs             611976       544    611432   1% /run
> udev               10240         0     10240   0% /dev
> tmpfs            1223952        76   1223876   1% /run/shm
> /dev/md2       292833568     33312 292800256   1% /or
> /dev/md1       292833568 202533276  90300292  70% /root/workplace
> /dev/sdc1          90297     52782     32698  62% /boot
> myon400:/      976694528 485310720 491383808  50% /srv
> /dev/sda8      137821156        28  90827392   1% /test
> 
> in this case, you can see a raid0 setup for meta and data, also
> acknowledged by "sdb8 id2" and "sdc8 id3", the overall amount
> of 1k blocks in df = 137821156 but the available amount is 
> only 90827392 1-k blocks while 28 are in use. this looks like
> the calculation of raid0 outputs something i would assume 
> for raid5 or in other words just 2 of 3 raid-members usable.
> 
> maybe i misinterpreted something or made a mistake by creation
> however any documentation around gave me no idea.
> 
> this test was made on a debian sid, using btrfs v0.19.

Could you try this patch? Maybe it can make you comfortable.

Subject: [PATCH] Btrfs: fix wrong inaccurate available space on raid0 profile

When we use raid0 as the data profile, df command may show us a very
inaccurate value of the available space, which may be much less than the
real one. It may make the users puzzled. Fix it by changing the calculation
of the available space, and making it be more similar to a fake chunk
allocation.

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
---
 fs/btrfs/super.c |   19 +++++++++++++------
 1 files changed, 13 insertions(+), 6 deletions(-)

diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index 8bd9d6d..01de53a 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -1079,7 +1079,7 @@ static int btrfs_calc_avail_data_space(struct btrfs_root *root, u64 *free_bytes)
 	u64 avail_space;
 	u64 used_space;
 	u64 min_stripe_size;
-	int min_stripes = 1;
+	int min_stripes = 1, num_stripes = 1;
 	int i = 0, nr_devices;
 	int ret;
 
@@ -1093,12 +1093,16 @@ static int btrfs_calc_avail_data_space(struct btrfs_root *root, u64 *free_bytes)
 
 	/* calc min stripe number for data space alloction */
 	type = btrfs_get_alloc_profile(root, 1);
-	if (type & BTRFS_BLOCK_GROUP_RAID0)
+	if (type & BTRFS_BLOCK_GROUP_RAID0) {
 		min_stripes = 2;
-	else if (type & BTRFS_BLOCK_GROUP_RAID1)
+		num_stripes = nr_devices;
+	} else if (type & BTRFS_BLOCK_GROUP_RAID1) {
 		min_stripes = 2;
-	else if (type & BTRFS_BLOCK_GROUP_RAID10)
+		num_stripes = 2;
+	} else if (type & BTRFS_BLOCK_GROUP_RAID10) {
 		min_stripes = 4;
+		num_stripes = 4;
+	}
 
 	if (type & BTRFS_BLOCK_GROUP_DUP)
 		min_stripe_size = 2 * BTRFS_STRIPE_LEN;
@@ -1167,13 +1171,16 @@ static int btrfs_calc_avail_data_space(struct btrfs_root *root, u64 *free_bytes)
 	i = nr_devices - 1;
 	avail_space = 0;
 	while (nr_devices >= min_stripes) {
+		if (num_stripes > nr_devices)
+			num_stripes = nr_devices;
+
 		if (devices_info[i].max_avail >= min_stripe_size) {
 			int j;
 			u64 alloc_size;
 
-			avail_space += devices_info[i].max_avail * min_stripes;
+			avail_space += devices_info[i].max_avail * num_stripes;
 			alloc_size = devices_info[i].max_avail;
-			for (j = i + 1 - min_stripes; j <= i; j++)
+			for (j = i + 1 - num_stripes; j <= i; j++)
 				devices_info[j].max_avail -= alloc_size;
 		}
 		i--;
-- 
1.7.6.4