All of lore.kernel.org
 help / color / mirror / Atom feed
* reiserfs tails and disk space
@ 2006-01-17 18:15 Bruce Guenter
  2006-01-17 18:58 ` Hans Reiser
  0 siblings, 1 reply; 6+ messages in thread
From: Bruce Guenter @ 2006-01-17 18:15 UTC (permalink / raw)
  To: reiserfs-list

[-- Attachment #1: Type: text/plain, Size: 1740 bytes --]

Hi.

I've been running a few tests with reiserfs and tails, and have been
unable to create a setup where the use (or lack) of tails results in a
significant difference in the amount of disk space used.

Here's what I've done:

1. Create a fresh 1GB filesystem (in a file on loopback), using reiserfs
with no options.

2. Mount the filesystem with either no options, "notail", "tails=off",
"tails=on", or "tails=small".

3. Unpack a sources tarball onto the filesystem, consisting of two fully
compiled versions of the linux kernel.  The tarball contains 47996 files
and 3321 directories totalling about 660MB of space.

4. Measure the free disk space using df.

5. Use dd to fill up the free disk space and count how many 1kB blocks
it could write.

In all of the tests, the result was within 12kB of each other.  In fact,
the tests with "notail" or "tails=off" options had more usable disk
space than when using tails.

Results:

Options    1K-blocks    Used Available
default      1023964  645988  377976
notail       1023964  645988  377976
tails=off    1023964  645996  377968
tails=on     1023964  646000  377964
tails=small  1023964  645996  377968

default      377600+0 records out
notail       377600+0 records out
tails=off    377592+0 records out
tails=on     377588+0 records out
tails=small  377592+0 records out

I've put the log files and scripts up for review at
	http://untroubled.org/reiserfsdf/
I'm using Gentoo Linux, kernel 2.6.14-gentoo-r5

Am I missing something, is this an expected result, or is something
broken?

Thanks.
-- 
Bruce Guenter <bruce@untroubled.org> http://untroubled.org/
OpenPGP key: 699980E8 / D0B7 C8DD 365D A395 29DA  2E2A E96F B2DC 6999 80E8

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: reiserfs tails and disk space
  2006-01-17 18:15 reiserfs tails and disk space Bruce Guenter
@ 2006-01-17 18:58 ` Hans Reiser
  2006-01-18 18:34   ` Vitaly Fertman
  0 siblings, 1 reply; 6+ messages in thread
From: Hans Reiser @ 2006-01-17 18:58 UTC (permalink / raw)
  To: Bruce Guenter; +Cc: reiserfs-list, vitaly

The result is not expected, Vitaly please look into it.

Hans

Bruce Guenter wrote:

>Hi.
>
>I've been running a few tests with reiserfs and tails, and have been
>unable to create a setup where the use (or lack) of tails results in a
>significant difference in the amount of disk space used.
>
>Here's what I've done:
>
>1. Create a fresh 1GB filesystem (in a file on loopback), using reiserfs
>with no options.
>
>2. Mount the filesystem with either no options, "notail", "tails=off",
>"tails=on", or "tails=small".
>
>3. Unpack a sources tarball onto the filesystem, consisting of two fully
>compiled versions of the linux kernel.  The tarball contains 47996 files
>and 3321 directories totalling about 660MB of space.
>
>4. Measure the free disk space using df.
>
>5. Use dd to fill up the free disk space and count how many 1kB blocks
>it could write.
>
>In all of the tests, the result was within 12kB of each other.  In fact,
>the tests with "notail" or "tails=off" options had more usable disk
>space than when using tails.
>
>Results:
>
>Options    1K-blocks    Used Available
>default      1023964  645988  377976
>notail       1023964  645988  377976
>tails=off    1023964  645996  377968
>tails=on     1023964  646000  377964
>tails=small  1023964  645996  377968
>
>default      377600+0 records out
>notail       377600+0 records out
>tails=off    377592+0 records out
>tails=on     377588+0 records out
>tails=small  377592+0 records out
>
>I've put the log files and scripts up for review at
>	http://untroubled.org/reiserfsdf/
>I'm using Gentoo Linux, kernel 2.6.14-gentoo-r5
>
>Am I missing something, is this an expected result, or is something
>broken?
>
>Thanks.
>  
>


^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: reiserfs tails and disk space
  2006-01-17 18:58 ` Hans Reiser
@ 2006-01-18 18:34   ` Vitaly Fertman
  2006-01-19 19:53     ` Bruce Guenter
  0 siblings, 1 reply; 6+ messages in thread
From: Vitaly Fertman @ 2006-01-18 18:34 UTC (permalink / raw)
  To: Bruce Guenter; +Cc: Hans Reiser, reiserfs-list

[-- Attachment #1: Type: text/plain, Size: 1975 bytes --]

On Tuesday 17 January 2006 21:58, Hans Reiser wrote:
> The result is not expected, Vitaly please look into it.
> 
> Hans
> 
> Bruce Guenter wrote:
> 
> >Hi.
> >
> >I've been running a few tests with reiserfs and tails, and have been
> >unable to create a setup where the use (or lack) of tails results in a
> >significant difference in the amount of disk space used.

thank you for the report, the attached patch should fix 
the broken mount options. please try it.

> >Here's what I've done:
> >
> >1. Create a fresh 1GB filesystem (in a file on loopback), using reiserfs
> >with no options.
> >
> >2. Mount the filesystem with either no options, "notail", "tails=off",
> >"tails=on", or "tails=small".
> >
> >3. Unpack a sources tarball onto the filesystem, consisting of two fully
> >compiled versions of the linux kernel.  The tarball contains 47996 files
> >and 3321 directories totalling about 660MB of space.
> >
> >4. Measure the free disk space using df.
> >
> >5. Use dd to fill up the free disk space and count how many 1kB blocks
> >it could write.
> >
> >In all of the tests, the result was within 12kB of each other.  In fact,
> >the tests with "notail" or "tails=off" options had more usable disk
> >space than when using tails.
> >
> >Results:
> >
> >Options    1K-blocks    Used Available
> >default      1023964  645988  377976
> >notail       1023964  645988  377976
> >tails=off    1023964  645996  377968
> >tails=on     1023964  646000  377964
> >tails=small  1023964  645996  377968
> >
> >default      377600+0 records out
> >notail       377600+0 records out
> >tails=off    377592+0 records out
> >tails=on     377588+0 records out
> >tails=small  377592+0 records out
> >
> >I've put the log files and scripts up for review at
> >	http://untroubled.org/reiserfsdf/
> >I'm using Gentoo Linux, kernel 2.6.14-gentoo-r5
> >
> >Am I missing something, is this an expected result, or is something
> >broken?
> >
> >Thanks.
> >  
> >
> 
> 
> 

-- 
Vitaly

[-- Attachment #2: linux-2.6.14-reiserfs-mount-options-fix.patch --]
[-- Type: text/x-diff, Size: 436 bytes --]

--- linux-2.6.15-rc5-mm3-clean/fs/reiserfs/super.c	2005-12-21 23:57:54.000000000 +0300
+++ linux-2.6.15-rc5-mm3/fs/reiserfs/super.c	2006-01-18 21:28:25.206460792 +0300
@@ -1131,7 +1131,7 @@
 			REISERFS_SB(s)->s_mount_opt &= ~(1 << REISERFS_ATTRS);
 		}
 	} else if (le32_to_cpu(rs->s_flags) & reiserfs_attrs_cleared) {
-		REISERFS_SB(s)->s_mount_opt |= REISERFS_ATTRS;
+		REISERFS_SB(s)->s_mount_opt |= (1 << REISERFS_ATTRS);
 	}
 }
 

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: reiserfs tails and disk space
  2006-01-18 18:34   ` Vitaly Fertman
@ 2006-01-19 19:53     ` Bruce Guenter
  2006-01-19 22:00       ` Vitaly Fertman
  0 siblings, 1 reply; 6+ messages in thread
From: Bruce Guenter @ 2006-01-19 19:53 UTC (permalink / raw)
  To: reiserfs-list

[-- Attachment #1: Type: text/plain, Size: 487 bytes --]

On Wed, Jan 18, 2006 at 09:34:49PM +0300, Vitaly Fertman wrote:
> thank you for the report, the attached patch should fix 
> the broken mount options. please try it.

It does indeed fix the problem.  What other mount options would have
been affected by this problem?  In what version of the kernel was this
problem introduced?

Thanks.
-- 
Bruce Guenter <bruce@untroubled.org> http://untroubled.org/
OpenPGP key: 699980E8 / D0B7 C8DD 365D A395 29DA  2E2A E96F B2DC 6999 80E8

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: reiserfs tails and disk space
  2006-01-19 19:53     ` Bruce Guenter
@ 2006-01-19 22:00       ` Vitaly Fertman
  2006-01-21 18:08         ` Laurent Riffard
  0 siblings, 1 reply; 6+ messages in thread
From: Vitaly Fertman @ 2006-01-19 22:00 UTC (permalink / raw)
  To: reiserfs-list; +Cc: Bruce Guenter

On Thursday 19 January 2006 22:53, Bruce Guenter wrote:
> On Wed, Jan 18, 2006 at 09:34:49PM +0300, Vitaly Fertman wrote:
> > thank you for the report, the attached patch should fix 
> > the broken mount options. please try it.
> 
> It does indeed fix the problem.  What other mount options would have
> been affected by this problem?  

it does not seem to brake anything else.

> In what version of the kernel was this 
> problem introduced?

2.6.13.

-- 
Vitaly

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: reiserfs tails and disk space
  2006-01-19 22:00       ` Vitaly Fertman
@ 2006-01-21 18:08         ` Laurent Riffard
  0 siblings, 0 replies; 6+ messages in thread
From: Laurent Riffard @ 2006-01-21 18:08 UTC (permalink / raw)
  To: reiserfs-list; +Cc: Bruce Guenter

Le 19.01.2006 23:00, Vitaly Fertman a écrit :
> On Thursday 19 January 2006 22:53, Bruce Guenter wrote:
> 
>>On Wed, Jan 18, 2006 at 09:34:49PM +0300, Vitaly Fertman wrote:
>>
>>>thank you for the report, the attached patch should fix 
>>>the broken mount options. please try it.
>>
>>It does indeed fix the problem.  What other mount options would have
>>been affected by this problem?  
> 
> 
> it does not seem to brake anything else.
> 
> 
>>In what version of the kernel was this 
>>problem introduced?
> 
> 
> 2.6.13.
> 

This patch have been pushed into 2.6.16-rc1-mm2, and now I have some
permissions problems : I'm not able to update some files or
directories althought the permissions seems OK. Not all files are
impacted.

Reverting
someone-broke-reiserfs-v3-mount-options-this-fixes-it.patch fixes my
permissions problems.

I have to say this filesystem have been badly hit by the
2.6.16-rc1-mm1 reiserfs' bugs. I had to issue a "fsck.reiserfs
--rebuild-tree" to get back about 1 Gb of files in lost+found
(reiserfsck 3.6.19).

Here are some logs taken with an unpatched 2.6.16-rc1-mm2 kernel :

Script started on Sat Jan 21 17:54:22 2006
[ this is root user ]
sh-3.00# cd /home/gnutella/
sh-3.00# df -h .
Filesystem            Size  Used Avail Use% Mounted on
/dev/mapper/vglinux1-lvhome
                      5.0G  4.8G  291M  95% /home
sh-3.00# mount
[...]
/dev/mapper/vglinux1-lvhome on /home type reiserfs (rw)
[...]
sh-3.00# ls -ld .
drwx--x--x  18 gnutella gnutella 768 Jan 21 16:35 .
sh-3.00# ls -l
total 16
drwxr-xr-x  3 gnutella gnutella  72 Dec 24  2002 News
lrwxrwxrwx  1 gnutella gnutella  21 Jan 20 00:19 azureus ->
/home/laurent/azureus
-rw-r--r--  1 root     root     121 Jan 21 16:35 essai
-rw-r--r--  1 root     root     121 Jan 21 15:44 essai2
-rw-r--r--  1 root     root     121 Jan 21 16:24 essai3
-rw-r--r--  1 root     root       0 Jan 21 16:35 essai4
drwxr-xr-x  5 gnutella gnutella 120 Jan 20 00:11 klibido
-rw-------  1 gnutella gnutella 377 Jan 20 00:22 nohup.out
drwx------  7 gnutella gnutella 320 Jan 21 04:08 tmp
sh-3.00# touch essai5
touch: cannot touch `essai5': Permission denied
sh-3.00# > eesai5
sh: eesai5: Permission denied
sh-3.00# rm essai
rm: remove write-protected regular file `essai'? y
rm: cannot remove `essai': Permission denied
sh-3.00# strace touch essai
execve("/bin/touch", ["touch", "essai"], [/* 13 vars */]) = 0
uname({sys="Linux", node="antares.localdomain", ...}) = 0
brk(0)                                  = 0x8051000
[...]
open("essai", O_WRONLY|O_NONBLOCK|O_CREAT|O_NOCTTY|O_LARGEFILE,
0666) = -1 EACCES (Permission denied)
utime("essai", NULL)                    = -1 EACCES (Permission denied)
write(2, "touch: ", 7touch: )                  = 7
write(2, "cannot touch `essai\'", 20cannot touch `essai')   = 20
write(2, ": Permission denied", 19: Permission denied)     = 19
write(2, "\n", 1
)                       = 1
exit_group(1)                           = ?
Process 1699 detached
sh-3.00# ls -l
total 16
drwxr-xr-x  3 gnutella gnutella  72 Dec 24  2002 News
lrwxrwxrwx  1 gnutella gnutella  21 Jan 20 00:19 azureus ->
/home/laurent/azureus
-rw-r--r--  1 root     root     121 Jan 21 16:35 essai
-rw-r--r--  1 root     root     121 Jan 21 15:44 essai2
-rw-r--r--  1 root     root     121 Jan 21 16:24 essai3
-rw-r--r--  1 root     root       0 Jan 21 16:35 essai4
drwxr-xr-x  5 gnutella gnutella 120 Jan 20 00:11 klibido
-rw-------  1 gnutella gnutella 377 Jan 20 00:22 nohup.out
drwx------  7 gnutella gnutella 320 Jan 21 04:08 tmp
sh-3.00# strace vi essai 2>/tmp/trace
[...]
sh-3.00# grep essai /tmp/essai
execve("/bin/vi", ["vi", "essai"], [/* 13 vars */]) = 0
stat64("essai", {st_mode=S_IFREG|0644, st_size=121, ...}) = 0
stat64("/home/gnutella/essai", {st_mode=S_IFREG|0644, st_size=121,
...}) = 0
stat64("essai", {st_mode=S_IFREG|0644, st_size=121, ...}) = 0
stat64("essai", {st_mode=S_IFREG|0644, st_size=121, ...}) = 0
access("essai", W_OK)                   = -1 EACCES (Permission denied)
open("essai", O_RDONLY|O_LARGEFILE)     = 3
open("essai", O_RDONLY|O_LARGEFILE)     = 3
stat64("essai", {st_mode=S_IFREG|0644, st_size=121, ...}) = 0
sh-3.00#
Script done on Sat Jan 21 18:24:23 2006

~~
laurent


^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2006-01-21 18:08 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2006-01-17 18:15 reiserfs tails and disk space Bruce Guenter
2006-01-17 18:58 ` Hans Reiser
2006-01-18 18:34   ` Vitaly Fertman
2006-01-19 19:53     ` Bruce Guenter
2006-01-19 22:00       ` Vitaly Fertman
2006-01-21 18:08         ` Laurent Riffard

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.