[linux-lvm] Re: [PATCH] 2.4.25--pre4 VFS lvm_snapshots

[linux-lvm] Re: [PATCH] 2.4.25--pre4 VFS lvm_snapshots

[Christoph Hellwig]

On Wed, Jan 07, 2004 at 04:51:49PM +0100, Xose Vazquez Perez wrote:
> hi,
> 
> This patch is necessary to be able to mount snapshots of journalling
> filesystems. It was flying around lvm for long time, years!!.
> And LiNUX distributions bring it, Red Hat at least since 7.x.
> So, it should be sure.

The patch still doesn't get better by that.  And even if it was this
would clearly be a feature for 2.6, not 2.4.

Re: [linux-lvm] Re: [PATCH] 2.4.25--pre4 VFS lvm_snapshots

[Greg Freemyer]

On Wed, 2004-01-07 at 11:12, Christoph Hellwig wrote:
> On Wed, Jan 07, 2004 at 04:51:49PM +0100, Xose Vazquez Perez wrote:
> > hi,
> > 
> > This patch is necessary to be able to mount snapshots of journalling
> > filesystems. It was flying around lvm for long time, years!!.
> > And LiNUX distributions bring it, Red Hat at least since 7.x.
> > So, it should be sure.
> 
> The patch still doesn't get better by that.  And even if it was this
> would clearly be a feature for 2.6, not 2.4.
> 
I have not started looking at 2.6 yet, but somehow I assumed it would
already have the automatic freeze invocation logic in it.

Too bad that it doesn't, but with 2.4 and XFS I initiate the freeze from
user space and don't depend on the patch being installed.  

I assume all of the various filesystems have a way to invoke the freeze
from userspace.  If so, this patch is a nicety, not a neccesity.

Greg
-- 
Greg Freemyer

[linux-lvm] Re: [PATCH] 2.4.25--pre4 VFS lvm_snapshots

[Xose Vazquez Perez]

Christoph Hellwig wrote:

> The patch still doesn't get better by that.  And even if it was this
> would clearly be a feature for 2.6, not 2.4.

Without this patch there is no warranty that a LVM_snapshot is updated
like the FS that is trying to replicate. DANGER !!!

I only see one solution:
to stop FS IO, and then # sync; sync; sync; lvcreate -s ...

Maybe it's needed to write, in the code or a doc, some warning about this
or to put a pointer to this patch :-?

--
Software is like sex, it's better when it's bug free.

Re: [PATCH] 2.4.25--pre4 VFS lvm_snapshots

[Xose Vazquez Perez]

Christoph Hellwig wrote:

> The patch still doesn't get better by that.  And even if it was this
> would clearly be a feature for 2.6, not 2.4.

Without this patch there is no warranty that a LVM_snapshot is updated
like the FS that is trying to replicate. DANGER !!!

I only see one solution:
to stop FS IO, and then # sync; sync; sync; lvcreate -s ...

Maybe it's needed to write, in the code or a doc, some warning about this
or to put a pointer to this patch :-?

--
Software is like sex, it's better when it's bug free.