[linux-lvm] summary of what I need to do to mount snapshot on XFS

[linux-lvm] summary of what I need to do to mount snapshot on XFS

[Jeremy Hansen]

I've seen a lot of posts that people do have this working.  I was hoping
someone could give a good summary for the steps involved.  The goal is
being able to mount the snapshots volume while the live filesystem is
mounted.  I see bits and pieces of the answer in various posts, so could
someone break it down for me in a single mail?

Thanks
-jeremy

Re: [linux-lvm] summary of what I need to do to mount snapshot on XFS

[Adrian Head]

The basic concept is you create a snapshot LV
lvcreate -L1G -s -n snap /dev/vg/lv1

This will create a 1G snapshot logical volume called snap for the original 
logical volume /dev/vg/lv1

If this has completed successfully then you can mount it.

mount -t xfs -o nouuid /dev/vg/snap /mnt/snapshot

This will mount an xfs snapshot.  The -o nouuid is needed so that mount will 
not complain and refuse to mount an xfs volume with identical uuid's

This is a good tutorial:
http://www.sistina.com/lvm_howtos/lvm_howto/Common_tasks.html

Is this what you are after or are you having problems with lvcreate 
segfaulting and kernel Oopes?

On Wed, 9 Jan 2002 01:24, Jeremy Hansen wrote:
> I've seen a lot of posts that people do have this working.  I was hoping
> someone could give a good summary for the steps involved.  The goal is
> being able to mount the snapshots volume while the live filesystem is
> mounted.  I see bits and pieces of the answer in various posts, so could
> someone break it down for me in a single mail?
>
> Thanks
> -jeremy
>
>
>
>
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

-- 
Adrian Head

(Public Key available on request.)

Re: [linux-lvm] summary of what I need to do to mount snapshot on XFS

[Patrick Caulfield]

On Wed, Jan 09, 2002 at 10:06:04PM +1000, Adrian Head wrote:
> The basic concept is you create a snapshot LV
> lvcreate -L1G -s -n snap /dev/vg/lv1
> 
> This will create a 1G snapshot logical volume called snap for the original 
> logical volume /dev/vg/lv1
> 
> If this has completed successfully then you can mount it.
> 
> mount -t xfs -o nouuid /dev/vg/snap /mnt/snapshot

You may also need -o norecovery too.

patrick

Re: [linux-lvm] summary of what I need to do to mount snapshot on XFS

[Jeremy Hansen]

My issue wasn't so much with how to create the snapshot but how to
simultaneously mount the snapshot with the live filesystem.  I saw a big
thread about having to uncomment from lvm.c, did that but then got
unresolved symbols...

It seemed as if there were some issues and a large discussion related to
doing this with XFS.

Thanks
-jeremy

On Wed, 2002-01-09 at 05:02, Patrick Caulfield wrote:
> On Wed, Jan 09, 2002 at 10:06:04PM +1000, Adrian Head wrote:
> > The basic concept is you create a snapshot LV
> > lvcreate -L1G -s -n snap /dev/vg/lv1
> > 
> > This will create a 1G snapshot logical volume called snap for the original 
> > logical volume /dev/vg/lv1
> > 
> > If this has completed successfully then you can mount it.
> > 
> > mount -t xfs -o nouuid /dev/vg/snap /mnt/snapshot
> 
> You may also need -o norecovery too.
> 
> patrick
> 
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

Re: [linux-lvm] summary of what I need to do to mount snapshot on XFS

[Jeremy Hansen]

On Wed, 2002-01-09 at 05:02, Patrick Caulfield wrote:
> On Wed, Jan 09, 2002 at 10:06:04PM +1000, Adrian Head wrote:
> > The basic concept is you create a snapshot LV
> > lvcreate -L1G -s -n snap /dev/vg/lv1
> > 
> > This will create a 1G snapshot logical volume called snap for the original 
> > logical volume /dev/vg/lv1
> > 
> > If this has completed successfully then you can mount it.
> > 
> > mount -t xfs -o nouuid /dev/vg/snap /mnt/snapshot
> 
> You may also need -o norecovery too.

Ahh, it looks as if this did the trick.  I was able to mount it!

-jeremy
 
> patrick
> 
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

Re: [linux-lvm] summary of what I need to do to mount snapshot on XFS

[Jeremy Hansen]

Actually, now I am getting Oops

Mounting filesystem "lvm(58,1)" in no-recovery mode.  Filesystem will be
inconsi
stent.
lvm -- giving up to snapshot /dev/collector/collector_lv on
/dev/collector/colle
ctor_snapshot: out of space
Unable to handle kernel paging request at virtual address 00006e60
 printing eip:
f884898d
*pde = 00000000
Oops: 0000
CPU:    0
EIP:    0010:[<f884898d>]    Not tainted
EFLAGS: 00010246
eax: 00000dcc   ebx: 00006e60   ecx: c7612c00   edx: 00000000
esi: 00000000   edi: 022ae678   ebp: 00000800   esp: e5149a9c
ds: 0018   es: 0018   ss: 0018
Process mhonarc (pid: 21751, stackpage=e5149000)
Stack: 00000080 e5148000 00000027 00000000 c7612d70 c7612c00 022ae69f
00000003 
       f8845821 e5149af6 e5149aec 022ae178 c7612c00 00002202 c69e5360
00000f67 
       f7e90600 f785b000 022ae178 00011560 022ae69f 022ae69f 08000001
c69e5360 
Call Trace: [<f8845821>] [<c01b1ddf>] [<f884590f>] [<c0238f4f>]
[<c01b1ddf>] 
   [<c01b2007>] [<c01b24a0>] [<c01b211e>] [<c01b1e50>] [<c01f92e4>]
[<c01f9a27>]
 
   [<c020551e>] [<c01f9157>] [<c01fafe8>] [<c01fb327>] [<c01f8b59>]
[<c0204fd3>]
 
   [<c0217ffe>] [<c021a437>] [<c01cfa77>] [<c01f4044>] [<c020b2f3>]
[<c0219cd1>]
 
   [<c0149b9c>] [<c0147496>] [<c0142025>] [<c014204b>] [<c013f01e>]
[<c010712b>]
 

Code: 8b 0b eb 12 eb 0d 90 90 90 90 90 90 90 90 90 90 90 90 90 8b 


On Wed, 2002-01-09 at 05:02, Patrick Caulfield wrote:
> On Wed, Jan 09, 2002 at 10:06:04PM +1000, Adrian Head wrote:
> > The basic concept is you create a snapshot LV
> > lvcreate -L1G -s -n snap /dev/vg/lv1
> > 
> > This will create a 1G snapshot logical volume called snap for the original 
> > logical volume /dev/vg/lv1
> > 
> > If this has completed successfully then you can mount it.
> > 
> > mount -t xfs -o nouuid /dev/vg/snap /mnt/snapshot
> 
> You may also need -o norecovery too.
> 
> patrick
> 
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

Re: [linux-lvm] summary of what I need to do to mount snapshot on XFS

[Adrian Head]

If this works for you - good.  I have had heaps of issues that I have been 
working with various developers and trying to get fixed.

Basically if I used the SGI 2.4.17-xfs kernel with lvm-1.0.1rc4(ish) some 
file systems would work but others would fail.  Some would fail on snapshot 
creation, others would fail on snapshot mount and others would fail when the 
snapshot overflowed.  When I upgraded LVM to lvm-1.0.1 other filesystems 
issues were fixed but others broke again.  Check out the "Unable to get XFS, 
ext3, reiserfs & LVM to coexist happily"
http://marc.theaimsgroup.com/?l=linux-lvm&m=100994364215110&w=2
 thread on linux-lvm / linux-xfs mail archives and see if any of this relates 
to your problem.  With 2.4.17-xfs + lvm-1.0.1 upgrade patch + VFS-lock patch 
ext2/3 & resierfs will operate correctly but XFS will fail at snapshot 
creation.  You can do a little source code change and get a XFS snapshot 
created and mounted; but then it will fail on an overfull snapshot.  These 
problems are being worked on by myself and the great SGI XFS developers.

The lvm.c change you talk about I assume is the LVM_VFS_ENHANCEMENT change?  
For some reason when adding the lvm-1.0.1 upgrade patch this disappears from 
lvm.c   I have been able to get this to compile with the change but I have 
forgoten what I did.  

This is one question I would like answered by the LVM gurus:  should #define 
LVM_VFS_ENHANCEMENT still be there after upgrading to lvm-1.0.1?  What does 
LVM_VFS_ENHANCEMENT actually do w.r.t lvm-1.0.1rc4(ish) and lvm-1.0.1?  I 
know that LVM_VFS_ENHANCEMENT in older LVM versions was the locking of the 
VFS so that snapshots would be consistant - is this still correct and is it 
still needed?

For the "large discussion" you are talking about the "Unable to get XFS, 
ext3, reiserfs & LVM to coexist happily" thread - hit me off list and I will 
attempt to summarise it for you.

On Thu, 10 Jan 2002 06:09, Jeremy Hansen wrote:
> My issue wasn't so much with how to create the snapshot but how to
> simultaneously mount the snapshot with the live filesystem.  I saw a big
> thread about having to uncomment from lvm.c, did that but then got
> unresolved symbols...
>
> It seemed as if there were some issues and a large discussion related to
> doing this with XFS.
>
> Thanks
> -jeremy

-- 
Adrian Head

(Public Key available on request.)

Re: [linux-lvm] summary of what I need to do to mount snapshot on XFS

[Anselm Kruis]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 2122 bytes --]

On 9 Jan 2002, Jeremy Hansen wrote:

> On Wed, 2002-01-09 at 05:02, Patrick Caulfield wrote:
> > On Wed, Jan 09, 2002 at 10:06:04PM +1000, Adrian Head wrote:
> > > The basic concept is you create a snapshot LV
> > > lvcreate -L1G -s -n snap /dev/vg/lv1
> > > 
> > > This will create a 1G snapshot logical volume called snap for the original 
> > > logical volume /dev/vg/lv1
> > > 
> > > If this has completed successfully then you can mount it.
> > > 
> > > mount -t xfs -o nouuid /dev/vg/snap /mnt/snapshot
> > 
> > You may also need -o norecovery too.
> 
> Ahh, it looks as if this did the trick.  I was able to mount it!

Alternatively you could try the writeable snapshot patch. Then do an

lvchange -p rw /dev/vg/snap
mount -t xfs -o nouuid /dev/vg/snap /mnt/snapshot
umount /dev/vg/snap
sync
lvchange -p r /dev/vg/snap
# now the xfs on the snapshot is clean
mount -t xfs -o ro,nouuid /dev/vg/snap /mnt/snapshot

> 
> -jeremy
>  
> > patrick
> > 
> > 
> > _______________________________________________
> > linux-lvm mailing list
> > linux-lvm@sistina.com
> > http://lists.sistina.com/mailman/listinfo/linux-lvm
> > read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html
> 
> 
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html
> 

---
 Anselm Kruis                                 Tel. +49 (0)89-356386-74
 science + computing ag		              FAX  +49 (0)89-356386-37
 Ingolstädter Str. 22             mailto: A.Kruis@science-computing.de
 D-80807 München                 WWW: http://www.science-computing.de/

**********************************************************************
***     CeBIT 2002                                                 ***
***     Besuchen Sie uns auf der CeBIT vom 13.-20.03.2002          ***
***     auf dem Messegelände in Hannover, Halle 11                 ***
**********************************************************************

[-- Attachment #2: Type: TEXT/PLAIN, Size: 10926 bytes --]

diff -c -r -b -B -N -x CVS LVM.vanilla/PATCHES/linux-2.4.15-writeable-snapshots.patch LVM/PATCHES/linux-2.4.15-writeable-snapshots.patch
*** LVM.vanilla/PATCHES/linux-2.4.15-writeable-snapshots.patch	Thu Jan  1 01:00:00 1970
--- LVM/PATCHES/linux-2.4.15-writeable-snapshots.patch	Sun Nov 18 11:05:14 2001
***************
*** 0 ****
--- 1,27 ----
+ diff -u --recursive --new-file linux-2.4.15pre4.vanilla/Documentation/Configure.help linux/Documentation/Configure.help
+ --- linux-2.4.15pre4.vanilla/Documentation/Configure.help	Sun Nov 18 10:46:17 2001
+ +++ linux/Documentation/Configure.help	Sat Nov 17 22:55:13 2001
+ @@ -1693,6 +1693,13 @@
+    want), say M here and read <file:Documentation/modules.txt>.  The
+    module will be called lvm-mod.o.
+  
+ +Support for writeable snapshots
+ +CONFIG_LVM_WRITEABLE_SNAPSHOTS
+ +  The LVM driver allows you to make snapshots of a logocal volume.
+ +  Usually these snapshots are read-only. This options add support for
+ +  writeable snapshots. You can use a writeable snapshot to run a
+ +  filesystem check on the snapshot prior to mounting it.
+ +
+  Multiple devices driver support (RAID and LVM)
+  CONFIG_MD
+    Support multiple physical spindles through a single logical device.
+ diff -u --recursive --new-file linux-2.4.15pre4.vanilla/drivers/md/Config.in linux/drivers/md/Config.in
+ --- linux-2.4.15pre4.vanilla/drivers/md/Config.in	Fri Sep 14 23:22:18 2001
+ +++ linux/drivers/md/Config.in	Sat Nov 17 22:41:02 2001
+ @@ -14,5 +14,6 @@
+  dep_tristate '  Multipath I/O support' CONFIG_MD_MULTIPATH $CONFIG_BLK_DEV_MD
+  
+  dep_tristate ' Logical volume manager (LVM) support' CONFIG_BLK_DEV_LVM $CONFIG_MD
+ +dep_mbool     '  Support for writeable snapshots' CONFIG_LVM_WRITEABLE_SNAPSHOTS $CONFIG_BLK_DEV_LVM $CONFIG_EXPERIMENTAL
+  
+  endmenu
diff -c -r -b -B -N -x CVS LVM.vanilla/kernel/lvm.c LVM/kernel/lvm.c
*** LVM.vanilla/kernel/lvm.c	Tue Jan  1 21:49:48 2002
--- LVM/kernel/lvm.c	Tue Jan  1 22:04:01 2002
***************
*** 214,219 ****
--- 214,220 ----
   *                 in the LV each time.  [AED]
   *    12/10/2001 - Use add/del_gendisk() routines in 2.4.10+
   *    01/11/2001 - Backport read_ahead change from Linus kernel [AED]
+  *    01/01/2002 - Support for writeable snapshots [A.Kruis@science-computing.de]
   *
   */
  
***************
*** 374,379 ****
--- 375,385 ----
  ushort lvm_iop_version = LVM_DRIVER_IOP_VERSION;
  int loadtime = 0;
  const char *const lvm_name = LVM_NAME;
+ ushort lvm_feature_bits =
+ #ifdef CONFIG_LVM_WRITEABLE_SNAPSHOTS
+         LVM_FEATURE_WRITEABLE_SNAPSHOTS |
+ #endif
+ 0;
  
  
  /* volume group descriptor area pointers */
***************
*** 717,722 ****
--- 723,735 ----
  
  	/* Main command switch */
  	switch (command) {
+ 	case LVM_GET_FEATURE_BITS:
+ 		/* return a the feature bits. These bits indicate particular
+ 		   features supported by this kernel. */
+ 		if (copy_to_user(arg, &lvm_feature_bits, sizeof(ushort)) != 0)
+ 			return -EFAULT;
+ 		return 0;
+ 
  	case LVM_LOCK_LVM:
  		/* lock the LVM */
  		return lvm_do_lock_lvm();
***************
*** 1063,1068 ****
--- 1076,1084 ----
  			return -EFAULT;
  		break;
  
+ 	case BLKBSZGET:
+ 	case BLKBSZSET:
+ 		return blk_ioctl (inode->i_rdev, command, a);
  
  	case HDIO_GETGEO:
  		/* get disk geometry */
***************
*** 1293,1298 ****
--- 1309,1315 ----
  	lv_t *lv = vg_this->lv[LV_BLK(minor)];
  
  
+  restart:
  	down_read(&lv->lv_lock);
  	if (!(lv->lv_status & LV_ACTIVE)) {
  		printk(KERN_ALERT
***************
*** 1381,1389 ****
  		goto out;
  
  	if (lv->lv_access & LV_SNAPSHOT) { /* remap snapshot */
! 		if (lvm_snapshot_remap_block(&rdev_map, &rsector_map,
! 					     pe_start, lv) < 0)
  			goto bad;
  
  	} else if (rw == WRITE || rw == WRITEA) { /* snapshot origin */
  		lv_t *snap;
--- 1398,1425 ----
  		goto out;
  
  	if (lv->lv_access & LV_SNAPSHOT) { /* remap snapshot */
! 		int ret;
! 		ret = lvm_snapshot_remap_block(&rdev_map, &rsector_map,
! 						pe_start, lv);
! 		if (ret < 0) 
  			goto bad;
+ 		if (ret > 0 || !(rw == WRITE || rw == WRITEA))
+ 			goto out; /* already remapped or read access */
+ 
+ #ifdef CONFIG_LVM_WRITEABLE_SNAPSHOTS
+ 		/* remap the block */
+ 
+ 		/* Serializes the COW with the accesses to the
+ 		   snapshot device.
+ 
+ 		   __remap_snapshot acquires the write lock. */
+ 		up_read(&lv->lv_lock);
+ 		__remap_snapshot(rdev_map, rsector_map,
+ 				 pe_start, lv, vg_this);
+ 		goto restart;
+ #else
+ 		goto bad;
+ #endif
  
  	} else if (rw == WRITE || rw == WRITEA) { /* snapshot origin */
  		lv_t *snap;
diff -c -r -b -B -N -x CVS LVM.vanilla/kernel/lvm.h LVM/kernel/lvm.h
*** LVM.vanilla/kernel/lvm.h	Tue Jan  1 21:49:48 2002
--- LVM/kernel/lvm.h	Thu Nov 29 22:06:46 2001
***************
*** 72,77 ****
--- 72,78 ----
   *    22/06/2001 - added Andreas Dilger's PE on 4k boundary alignment enhancements
   *    19/07/2001 - added rwsem compatibility macros for 2.2 kernels
   *    13/11/2001 - reduced userspace inclusion of kernel headers to a minimum
+  *    18/11/2001 - added defines for LVM_GET_FEATURE_BITS ioctl [A.Kruis]
   *
   */
  
***************
*** 368,374 ****
--- 369,384 ----
  /* This is actually the same as _IO ( 0xff, 0x00), oops.  Remove for IOP 12+ */
  #define	LVM_LOCK_LVM            _IO ( 0xfe, 0x100)
  #endif
+ /* get lvm feature bits */
+ #define LVM_GET_FEATURE_BITS    _IOR ( 0xfe, 0x9b, 1)
  /* END ioctls */
+ 
+ 
+ /*
+  * Feature Bits
+  */
+ /* makeing a snapshot witeable is save */
+ #define LVM_FEATURE_WRITEABLE_SNAPSHOTS 0x01
  
  
  /*
diff -c -r -b -B -N -x CVS LVM.vanilla/tools/lib/liblvm.h LVM/tools/lib/liblvm.h
*** LVM.vanilla/tools/lib/liblvm.h	Tue Jan  1 21:49:48 2002
--- LVM/tools/lib/liblvm.h	Thu Nov 29 22:06:46 2001
***************
*** 409,414 ****
--- 409,415 ----
  void   lvm_dont_interrupt ( int);
  int    lvm_get_col_numbers ( char *, long **);
  int    lvm_get_iop_version ( void);
+ int    lvm_get_feature_bits ( void);
  void   lvm_init(int argc, char **argv);
  void   lvm_interrupt ( void);
  int    lvm_lock ( void);
diff -c -r -b -B -N -x CVS LVM.vanilla/tools/lib/lvm_get_feature_bits.c LVM/tools/lib/lvm_get_feature_bits.c
*** LVM.vanilla/tools/lib/lvm_get_feature_bits.c	Thu Jan  1 01:00:00 1970
--- LVM/tools/lib/lvm_get_feature_bits.c	Mon Nov 19 19:49:11 2001
***************
*** 0 ****
--- 1,60 ----
+ /*
+  * tools/lib/lvm_get_iop_version.c
+  *
+  * Copyright (C) 2001 Anselm Kruis
+  *
+  * November 2001
+  *
+  *
+  * This LVM library is free software; you can redistribute it and/or
+  * modify it under the terms of the GNU Library General Public
+  * License as published by the Free Software Foundation; either
+  * version 2 of the License, or (at your option) any later version.
+  *
+  * This LVM library is distributed in the hope that it will be useful,
+  * but WITHOUT ANY WARRANTY; without even the implied warranty of
+  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+  * Library General Public License for more details.
+  *
+  * You should have received a copy of the GNU Library General Public
+  * License along with this LVM library; if not, write to the Free
+  * Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+  * MA 02111-1307, USA
+  *
+  */
+ 
+ /*
+  * Changelog
+  *
+  *    18/11/2001 - created
+  *
+  */
+ 
+ #include <liblvm.h>
+ 
+ int lvm_get_feature_bits ( void) {
+    int group = -1;
+    int ret = 0;
+    ushort lvm_feature_bits = 0;
+ 
+    debug_enter ( "lvm_get_feature_bits -- CALLED\n");
+ 
+    /* just to ensure existance of /dev/lvm interface special */
+    lvm_check_special ();
+ 
+    if ( ( group = open ( LVM_DEV, O_RDONLY)) == -1)
+       ret = -LVM_ELVM_IOP_VERSION_OPEN;
+    else if ( ( ret = ioctl ( group, LVM_GET_FEATURE_BITS,
+                              &lvm_feature_bits)) == -1) ret = -errno;
+    debug ( "lvm_get_feature_bits -- AFTER ioctl ret: %d\n", ret);
+    if ( group != -1) close ( group);
+    /* older kernels didn't have this IOCTL */
+    if ( -ret == EINVAL ) {
+      ret = 0;
+      lvm_feature_bits = 0; /* no special features */
+    }
+    if ( ret == 0) ret = lvm_feature_bits;
+ 
+    debug_leave ( "lvm_get_feature_bits -- LEAVING with ret: %d\n", ret);
+    return ret;
+ }
diff -c -r -b -B -N -x CVS LVM.vanilla/tools/lvchange.c LVM/tools/lvchange.c
*** LVM.vanilla/tools/lvchange.c	Wed Jul 25 16:52:19 2001
--- LVM/tools/lvchange.c	Mon Nov 19 19:46:12 2001
***************
*** 323,334 ****
           continue;
        }
     
!       if ( lv->lv_access & LV_SNAPSHOT) {
           fprintf ( stderr, "%s -- change on snapshot logical "
                             "volume \"%s\" not allowed\n", cmd, lv_name);
           continue;
        }
  
        if ( lv->lv_access & LV_SNAPSHOT_ORG) {
           fprintf ( stderr, "%s -- change on logical volume \"%s\" under "
                             "snapshot not allowed\n", cmd, lv_name);
--- 323,345 ----
           continue;
        }
     
!       if ( ( lv->lv_access & LV_SNAPSHOT) &&
!            ( (opt_C) || (opt_a) || (opt_r))) {
           fprintf ( stderr, "%s -- change on snapshot logical "
                             "volume \"%s\" not allowed\n", cmd, lv_name);
           continue;
        }
  
+       if ( ( lv->lv_access & LV_SNAPSHOT) && (opt_p)) {
+ 	 int feature_bits = lvm_get_feature_bits();
+ 	 if ( feature_bits < 0 || !(feature_bits & LVM_FEATURE_WRITEABLE_SNAPSHOTS)) {
+ 	   fprintf ( stderr, "%s -- change on snapshot logical "
+ 		     "volume \"%s\" not allowed\n", cmd, lv_name);
+ 	   continue;
+ 	 }
+       }
+       
+       /* fixme: opt_p should be save */
        if ( lv->lv_access & LV_SNAPSHOT_ORG) {
           fprintf ( stderr, "%s -- change on logical volume \"%s\" under "
                             "snapshot not allowed\n", cmd, lv_name);
***************
*** 364,370 ****
     
           if ( doit > 0) {
              doit_sum++;
!             lv->lv_access = lv_access;
              if ( opt_v > 0) printf ( "%s -- changing access permissions "
                                       "of logical volume \"%s\"\n",
                                       cmd, lv_name);
--- 375,383 ----
     
           if ( doit > 0) {
              doit_sum++;
!             /* don't clear access flags except for LV_WRITE and LV_READ */
!             lv->lv_access &= ~( LV_READ | LV_WRITE ) | lv_access;
!             lv->lv_access |= lv_access;
              if ( opt_v > 0) printf ( "%s -- changing access permissions "
                                       "of logical volume \"%s\"\n",
                                       cmd, lv_name);

Re: [linux-lvm] summary of what I need to do to mount snapshot on XFS

[Heinz J . Mauelshagen]

On Thu, Jan 10, 2002 at 09:38:08AM +1000, Adrian Head wrote:
> If this works for you - good.  I have had heaps of issues that I have been 
> working with various developers and trying to get fixed.
> 
> Basically if I used the SGI 2.4.17-xfs kernel with lvm-1.0.1rc4(ish) some 
> file systems would work but others would fail.  Some would fail on snapshot 
> creation, others would fail on snapshot mount and others would fail when the 
> snapshot overflowed.  When I upgraded LVM to lvm-1.0.1 other filesystems 
> issues were fixed but others broke again.  Check out the "Unable to get XFS, 
> ext3, reiserfs & LVM to coexist happily"
> http://marc.theaimsgroup.com/?l=linux-lvm&m=100994364215110&w=2
>  thread on linux-lvm / linux-xfs mail archives and see if any of this relates 
> to your problem.  With 2.4.17-xfs + lvm-1.0.1 upgrade patch + VFS-lock patch 
> ext2/3 & resierfs will operate correctly but XFS will fail at snapshot 
> creation.  You can do a little source code change and get a XFS snapshot 
> created and mounted; but then it will fail on an overfull snapshot.  These 
> problems are being worked on by myself and the great SGI XFS developers.
> 
> The lvm.c change you talk about I assume is the LVM_VFS_ENHANCEMENT change?  
> For some reason when adding the lvm-1.0.1 upgrade patch this disappears from 
> lvm.c   I have been able to get this to compile with the change but I have 
> forgoten what I did.  
> 
> This is one question I would like answered by the LVM gurus:  should #define 
> LVM_VFS_ENHANCEMENT still be there after upgrading to lvm-1.0.1?

Yes.

> What does 
> LVM_VFS_ENHANCEMENT actually do w.r.t lvm-1.0.1rc4(ish) and lvm-1.0.1?

What it did before: compile the VFS lock/unlock calls in.

> I 
> know that LVM_VFS_ENHANCEMENT in older LVM versions was the locking of the 
> VFS so that snapshots would be consistant - is this still correct and is it 
> still needed?

Yes.

> 
> For the "large discussion" you are talking about the "Unable to get XFS, 
> ext3, reiserfs & LVM to coexist happily" thread - hit me off list and I will 
> attempt to summarise it for you.
> 
> On Thu, 10 Jan 2002 06:09, Jeremy Hansen wrote:
> > My issue wasn't so much with how to create the snapshot but how to
> > simultaneously mount the snapshot with the live filesystem.  I saw a big
> > thread about having to uncomment from lvm.c, did that but then got
> > unresolved symbols...
> >
> > It seemed as if there were some issues and a large discussion related to
> > doing this with XFS.
> >
> > Thanks
> > -jeremy
> 
> -- 
> Adrian Head
> 
> (Public Key available on request.)
> 
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm
> read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

-- 

Regards,
Heinz    -- The LVM Guy --

*** Software bugs are stupid.
    Nevertheless it needs not so stupid people to solve them ***

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Heinz Mauelshagen                                 Sistina Software Inc.
Senior Consultant/Developer                       Am Sonnenhang 11
                                                  56242 Marienrachdorf
                                                  Germany
Mauelshagen@Sistina.com                           +49 2626 141200
                                                       FAX 924446
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Re: [linux-lvm] summary of what I need to do to mount snapshot on XFS

[Adrian Head]

On Wed, 16 Jan 2002 00:16, Heinz J . Mauelshagen wrote:
> > The lvm.c change you talk about I assume is the LVM_VFS_ENHANCEMENT
> > change? For some reason when adding the lvm-1.0.1 upgrade patch this
> > disappears from lvm.c   I have been able to get this to compile with the
> > change but I have forgoten what I did.
> >
> > This is one question I would like answered by the LVM gurus:  should
> > #define LVM_VFS_ENHANCEMENT still be there after upgrading to lvm-1.0.1?
>
> Yes.

I have found that using the generated lvm-1.0.1 upgrade patch that the 
#define LVM_VFS_ENHANCEMENT gets removed.  If all I do is add it back in 
manually - do I need to worry about other areas of the code that also need 
touch-ups?  How would I tell?

Thanks

-- 
Adrian Head

(Public Key available on request.)

Re: [linux-lvm] summary of what I need to do to mount snapshot on XFS

[Andreas Dilger]

On Jan 16, 2002  01:09 +1000, Adrian Head wrote:
> On Wed, 16 Jan 2002 00:16, Heinz J . Mauelshagen wrote:
> > > The lvm.c change you talk about I assume is the LVM_VFS_ENHANCEMENT
> > > change? For some reason when adding the lvm-1.0.1 upgrade patch this
> > > disappears from lvm.c   I have been able to get this to compile with the
> > > change but I have forgoten what I did.
> > >
> > > This is one question I would like answered by the LVM gurus:  should
> > > #define LVM_VFS_ENHANCEMENT still be there after upgrading to lvm-1.0.1?
> >
> > Yes.
> 
> I have found that using the generated lvm-1.0.1 upgrade patch that the 
> #define LVM_VFS_ENHANCEMENT gets removed.  If all I do is add it back in 
> manually - do I need to worry about other areas of the code that also need 
> touch-ups?  How would I tell?

Well, when you generate a new LVM patch it will just do a "diff" between
the lvm.c file in the LVM/kernel directory and the linux/md/lvm.c file.
If the VFS-lock patch is applied to the kernel sources and not the
LVM/kernel sources, it will be removed by the diff, meaning you will
need to apply the VFS-lock patch to the kernel again.

If you find you are doing a lot of patching, you could apply the VFS-lock
patch to the LVM/kernel/lvm.c file, and then subsequent patches that you
create will not have this part removed.

Cheers, Andreas
--
Andreas Dilger
http://sourceforge.net/projects/ext2resize/
http://www-mddsp.enel.ucalgary.ca/People/adilger/