Mike McElroy | 1 Apr 04:39 2006
Picon

Re: Metadata on separate volume

Nathan Scott wrote:
> On Wed, Mar 29, 2006 at 08:38:26PM -0500, Mike McElroy wrote:
>   
>> I would like to know if it's possible to set up xfs with all the 
>> metadata residing one one volume and the data on another and, if not, 
>>     
>
> Its is possible, OOC why do you want to do that?
>   
OOC - out of coffee?
The realtime (data) volume is for fast access streaming data. The 
metadata would be on raid 5 storage to minimize the chance of losing it 
or corruption. The bitmap of the realtime volume and any metadata from 
that volume would be located on the raid volume. Does that make sense?
>   
>> would it be a major modification to to the code to do that.
>>     
>
> You may be able to achieve what you want with the realtime subvolume,
> and if so, no, its not a major modification at all. ;)
>
> cheers.
>
>   
I have found very little documentation on the realtime volume.

Thanks for responding,

Nathan Scott | 1 Apr 07:34 2006
Picon

Re: Could not allocate buf during xfs_fsr?

On Fri, Mar 31, 2006 at 06:54:34AM -0500, Justin Piszcz wrote:
> What version is this patch against?

Shouhld apply to any recent version of the fsr source, not much has
changed here for quite some time.

cheers.

--

-- 
Nathan

Nathan Scott | 1 Apr 08:02 2006
Picon

Re: Metadata on separate volume

On Fri, Mar 31, 2006 at 09:39:45PM -0500, Mike McElroy wrote:
> Nathan Scott wrote:
> > On Wed, Mar 29, 2006 at 08:38:26PM -0500, Mike McElroy wrote:
> >   
> >> I would like to know if it's possible to set up xfs with all the 
> >> metadata residing one one volume and the data on another and, if not, 
> >>     
> >
> > Its is possible, OOC why do you want to do that?
> >   
> OOC - out of coffee?

Heh.  That and/or curiosity.

> The realtime (data) volume is for fast access streaming data. The 
> metadata would be on raid 5 storage to minimize the chance of losing it 
> or corruption. The bitmap of the realtime volume and any metadata from 
> that volume would be located on the raid volume. Does that make sense?

Yep.  Sounds a good match.

> I have found very little documentation on the realtime volume.

Yeah, its a bit like that.  Theres some on xfs(5), some more in
mkfs.xfs(8) under -r.

cheers.

--

-- 
Nathan
(Continue reading)

Timothy Shimmin | 2 Apr 08:01 2006
Picon

Re: patch for building/installing xfs-cmds on 64 bit platforms

Timothy Shimmin wrote:

> Hi there,
>
> In November last year, Jan Derfinak, supplied a patch for 
> building/installing xfs-cmds on amd64
> but Christoph H. had problems with "uname --hardware -platform" on 
> some 64 bit machines.
> (http://oss.sgi.com/archives/linux-xfs/2004-11/msg00115.html)
>
> Looking around I decided to adopt a solution from gettext which has a 
> config macro which
> looks at $CC -print-search-dirs for lib64 to make the decision.
> I've attached a patch just for xfs-cmds/attr to this email.
> If this works for people on various 64 bit platforms,
> then I'll do likewise for the other xfs related userspace libraries.
>
> Thanks muchly,
> Tim.

This is an update for my earlier patch for all the packages:
acl, attr, dmapi, xfsprogs, xfsdump, xfstests

If no comments/problems, then I'll probably check this in shortly.

 > diffstat multilib.patch
 acl/Makefile                    |    5
 acl/aclocal.m4                  |  250 +++++++++++++---------
 acl/configure.in                |    6
 acl/include/builddefs.in        |   11 -
(Continue reading)

Gerard Neil | 3 Apr 05:55 2006

xfsdump and chattr +d : undocumented but really useful behaviour

I've been running a gentoo linux server (currently 2.6.15) using xfs
and xfsdump for a couple of years now.

I've just discovered that if the the no-dump (d) attribute is set on a
directory (using chattr +d), then any new directories/files created in
that directory automatically acquire the attribute too.

This is *really* useful when used in conjunction with xfsdump -e,
because it means I can easily exclude entire subtrees from dumps.

If this is really how the no-dump attribute is supposed to behave on
xfs, then I suggest documenting this in the xfsdump man page. The
current "excluding individual files" section is confusing with respect
to chattr +d, and certainly doesn't mention this useful behaviour. I
can't find it documented anywhere else, either...

Regards,

Gerard

Bill Kendall | 3 Apr 16:32 2006
Picon

Re: xfsdump and chattr +d : undocumented but really useful behaviour

Hi Gerard,

This is a popular question lately. The man page was updated
a couple months back, around version 2.2.34 or so.

Regards,
Bill

On 04/02/06 22:55, Gerard Neil wrote:
> I've been running a gentoo linux server (currently 2.6.15) using xfs
> and xfsdump for a couple of years now.
> 
> I've just discovered that if the the no-dump (d) attribute is set on a
> directory (using chattr +d), then any new directories/files created in
> that directory automatically acquire the attribute too.
> 
> This is *really* useful when used in conjunction with xfsdump -e,
> because it means I can easily exclude entire subtrees from dumps.
> 
> If this is really how the no-dump attribute is supposed to behave on
> xfs, then I suggest documenting this in the xfsdump man page. The
> current "excluding individual files" section is confusing with respect
> to chattr +d, and certainly doesn't mention this useful behaviour. I
> can't find it documented anywhere else, either...
> 
> Regards,
> 
> Gerard
> 

(Continue reading)

Timothy Shimmin | 4 Apr 06:50 2006
Picon

TAKE 951422 - libdir paths in xfs-cmds Makefiles are wrong for many 64bit platforms

fix lib64 installs

Date:  Tue Apr  4 14:51:15 AEST 2006
Workarea:  puffy.melbourne.sgi.com:/home/tes/isms/xfs-cmds
Inspected by:  nathans <at> sgi.com

The following file(s) were checked into:
  longdrop.melbourne.sgi.com:/isms/xfs-cmds/master-melb

Modid:  master-melb:xfs-cmds:25657a
dmapi/m4/multilib.m4 - 1.1 - new
http://oss.sgi.com/cgi-bin/cvsweb.cgi/xfs-cmds/dmapi/m4/multilib.m4
xfstests/m4/multilib.m4 - 1.1 - new
http://oss.sgi.com/cgi-bin/cvsweb.cgi/xfs-cmds/xfstests/m4/multilib.m4
xfsdump/m4/multilib.m4 - 1.1 - new
http://oss.sgi.com/cgi-bin/cvsweb.cgi/xfs-cmds/xfsdump/m4/multilib.m4
acl/m4/multilib.m4 - 1.1 - new
http://oss.sgi.com/cgi-bin/cvsweb.cgi/xfs-cmds/acl/m4/multilib.m4
xfsprogs/m4/multilib.m4 - 1.1 - new
http://oss.sgi.com/cgi-bin/cvsweb.cgi/xfs-cmds/xfsprogs/m4/multilib.m4
attr/m4/multilib.m4 - 1.1 - new
http://oss.sgi.com/cgi-bin/cvsweb.cgi/xfs-cmds/attr/m4/multilib.m4
acl/configure.in - 1.29 - changed
http://oss.sgi.com/cgi-bin/cvsweb.cgi/xfs-cmds/acl/configure.in.diff?r1=text&tr1=1.29&r2=text&tr2=1.28&f=h
acl/Makefile - 1.19 - changed
http://oss.sgi.com/cgi-bin/cvsweb.cgi/xfs-cmds/acl/Makefile.diff?r1=text&tr1=1.19&r2=text&tr2=1.18&f=h
acl/include/builddefs.in - 1.32 - changed
http://oss.sgi.com/cgi-bin/cvsweb.cgi/xfs-cmds/acl/include/builddefs.in.diff?r1=text&tr1=1.32&r2=text&tr2=1.31&f=h
attr/configure.in - 1.18 - changed
http://oss.sgi.com/cgi-bin/cvsweb.cgi/xfs-cmds/attr/configure.in.diff?r1=text&tr1=1.18&r2=text&tr2=1.17&f=h
(Continue reading)

Evgeniy Sharapov | 4 Apr 09:42 2006
Picon

XFS partition recovery

I have XFS /home partition (fortunately not /). I can not mount it because
of the error during reading the superblock.

After running xfs_repair -nLv /dev/hda7 I got

<------cut--------->
................................
        would clear inode number in entry at offset 2416...
entry "20060331122258" at block 2 offset 2448 in directory inode 16922661
references free inode 184775172
        would clear inode number in entry at offset 2448...
entry "20060331122336" at block 2 offset 2480 in directory inode 16922661
references free inode 201635562
        would clear inode number in entry at offset 2480...
entry "20060331122347" at block 2 offset 2512 in directory inode 16922661
references free inode 218356061
        would clear inode number in entry at offset 2512...
entry "20060331122411" at block 2 offset 2544 in directory inode 16922661
references free inode 235060608
        would clear inode number in entry at offset 2544...
entry "20060331122457" at block 2 offset 2576 in directory inode 16922661
references free inode 252033667
        would clear inode number in entry at offset 2576...
entry "20060331122757" at block 2 offset 2640 in directory inode 16922661
references free inode 17026280        would clear inode number in entry at
offset 2640...
entry "20060331122931" at block 2 offset 2672 in directory inode 16922661
references free inode 33982501        would clear inode number in entry at
offset 2672...
entry "20060331123023" at block 2 offset 2704 in directory inode 16922661
(Continue reading)

compix | 4 Apr 14:27 2006
Picon

טכנאי מחשבים בחינם

המחשב מקולקל ? עם וירוסים ? איטי?

אל תחכו שהוא ייהרס לגמרי הזמינו היום טכנאי שיציל לכם את
המחשב, ייעל את העבודה שלו ויזרז את מהירותו !
"קומפיקס פתרונות מחשוב" יתנו לכם את כל הפתרונות שהמחשב
שלכם צריך ובזול :   (מקבלים ויזה ותשלומים!)

-תיקון המחשב בזריזות ויעילות ייחודית בבית הלקוח .
-שירותי מעבדה, מכירת מחשבים במחירי סיטונאי!! שדרוג
מחשבים וחלקי מחשב.
-הקמת רשתות אלחוטיות ,ביתיות ושיתוף מדפסות .
-חיבור לאינטרנט במחירים הזולים מכולם כולל התקנה
ושירות !

התקשרו עכשיו לייעוץ חינם!!!
אם לא תתקשרו לא תבינו כמה הפסדתם!!!

077-6633677 שירות כלל ארצי
www.compix.co.il





Ralf Hildebrandt | 4 Apr 23:21 2006
Picon

2.6.16.1: XFS internal error xfs_btree_check_lblock at line 215 of file fs/xfs/xfs_btree.c

I was running xfs_fsr on our /home at night when this happened:
(Kernel 2.6.16.1)

Security Events
=-=-=-=-=-=-=-=
Apr  2 00:26:17 postamt kernel: xfs_force_shutdown(sda5,0x8) called from line 1032 of file
fs/xfs/xfs_trans.c.  Return address = 0xb1128274

System Events
=-=-=-=-=-=-=
Apr  2 00:17:10 postamt fsr[28019]: /home start inode=0
Apr  2 00:25:29 postamt fsr[28938]: /home start inode=0
Apr  2 00:26:17 postamt kernel: Filesystem "sda5": XFS internal error xfs_btree_check_lblock at line 215
of file fs/xfs/xfs_btree.c.  Caller 0xb10dba58
Apr  2 00:26:17 postamt kernel:  [xfs_btree_check_lblock+82/407] xfs_btree_check_lblock+0x52/0x197
Apr  2 00:26:17 postamt kernel:  [xfs_bmbt_lookup+636/1471] xfs_bmbt_lookup+0x27c/0x5bf
Apr  2 00:26:17 postamt kernel:  [xfs_bmbt_lookup+636/1471] xfs_bmbt_lookup+0x27c/0x5bf
Apr  2 00:26:17 postamt kernel:  [xfs_bmap_del_extent+491/2774] xfs_bmap_del_extent+0x1eb/0xad6
Apr  2 00:26:17 postamt kernel:  [xfs_mod_incore_sb_batch+60/134] xfs_mod_incore_sb_batch+0x3c/0x86
Apr  2 00:26:17 postamt kernel:  [kmem_zone_alloc+61/176] kmem_zone_alloc+0x3d/0xb0
Apr  2 00:26:17 postamt kernel:  [xfs_bmap_search_extents+117/279] xfs_bmap_search_extents+0x75/0x117
Apr  2 00:26:17 postamt kernel:  [xfs_bunmapi+1851/4334] xfs_bunmapi+0x73b/0x10ee
Apr  2 00:26:17 postamt kernel:  [xfs_trans_ijoin+43/134] xfs_trans_ijoin+0x2b/0x86
Apr  2 00:26:17 postamt kernel:  [xfs_itruncate_finish+617/942] xfs_itruncate_finish+0x269/0x3ae
Apr  2 00:26:17 postamt kernel:  [xfs_trans_ijoin+43/134] xfs_trans_ijoin+0x2b/0x86
Apr  2 00:26:17 postamt kernel:  [xfs_inactive+1058/1282] xfs_inactive+0x422/0x502
Apr  2 00:26:17 postamt fsr[28938]: could not remove tmpdir: /home/.fsr/ag0: Input/output error
Apr  2 00:26:17 postamt kernel:  [linvfs_clear_inode+140/164] linvfs_clear_inode+0x8c/0xa4
Apr  2 00:26:17 postamt kernel:  [dquot_drop+0/106] dquot_drop+0x0/0x6a
Apr  2 00:26:17 postamt kernel:  [clear_inode+173/290] clear_inode+0xad/0x122
(Continue reading)


Gmane