Subject: Re: Which snapshot strategy to use? was: How to capture all
To: Jason Thorpe <thorpej@wasabisystems.com>
From: Christian Limpach <chris@pin.lu>
List: tech-kern
Date: 10/23/2003 18:14:12
On Thu, 23 Oct 2003 08:33:28 -0700 Jason Thorpe <thorpej@wasabisystems.com> wrote:

> On Thursday, October 23, 2003, at 05:47  AM, Christian Limpach wrote:
> 
> >> Let's say you have /u1 mounted from /dev/sd1a.  How would you then 
> >> take
> >> a snapshot?
> >
> > The driver provides virtualization, so you'd use it with a userland 
> > volume
> > manager or with wedges and then every device you mount is provided 
> > through
> > the driver.  For devices mounted through the driver, you can ask the 
> > driver
> > to suspend access to the device (while keeping it mounted), replace the
> > mapping and then resume access to the device.
> 
> Your examples aren't using a file system mounted on /dev/sd1a.  They're 
> using a file system mounted on /dev/mapper/vgdelight-test.  In other 
> words, you provided an example that is not what I asked for, and proved 
> my point at the same time -- it requires a priori knowledge (and a 
> configuration change, and use of some other software) to do snapshots 
> with a pseudo-device shim.

You're right.  But it's only true because sd/wd/*d parse disklabels and
provide these partition devices without any flexibility.  I could rip the
partition code out of sd/wd/*d and provide the partitions through the
driver (and name them /dev/sd1a...).  I think you call that wedges around
here (+- reading/parsing the disklabel in userspace, I'd hope).

You're right that in the current system, it requires a priori knowledge.

-- 
Christian Limpach <chris@pin.lu>