Re: [linux-lvm] Why vgscan not in kernel?

Re: [linux-lvm] Why vgscan not in kernel?

[Heinz J. Mauelshagen]

Dirk,

that's already on the TODO list as a post 1.0 feature.

Regards,
Heinz    -- The LVM Guy --


On Mon, Jun 25, 2001 at 02:24:13PM +0200, Dirk Heinrichs wrote:
> Hi all,
> 
> upon boot, SCSI and IDE drivers check harddrives attached to the device
> for partitions. Wouldn't it be better to have similar functionality
> (scanning for VG's) for LVM in the kernel, instead of using a userland
> command?
> 
> Bye...
> 
> 	Dirk
> -- 
> Dirk Heinrichs		| Tel:	+49 (0)241 413 260
> Configuration Manager	| Fax:	+49 (0)241 413 2640
> QIS Systemhaus GmbH	| Mail:	dheinrichs@qis-systemhaus.de
> J�licher Str. 338b	| Web:	http://www.qis-systemhaus.de
> D-52070 Aachen		| ICQ#:	110037733
> _______________________________________________
> 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

*** 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
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

[linux-lvm] Why vgscan not in kernel?

[Dirk Heinrichs]

Hi all,

upon boot, SCSI and IDE drivers check harddrives attached to the device
for partitions. Wouldn't it be better to have similar functionality
(scanning for VG's) for LVM in the kernel, instead of using a userland
command?

Bye...

	Dirk
-- 
Dirk Heinrichs		| Tel:	+49 (0)241 413 260
Configuration Manager	| Fax:	+49 (0)241 413 2640
QIS Systemhaus GmbH	| Mail:	dheinrichs@qis-systemhaus.de
J�licher Str. 338b	| Web:	http://www.qis-systemhaus.de
D-52070 Aachen		| ICQ#:	110037733

Re: [linux-lvm] Why vgscan not in kernel?

[Steven Lembark]

>> upon boot, SCSI and IDE drivers check harddrives attached to the device
>> for partitions. Wouldn't it be better to have similar functionality
>> (scanning for VG's) for LVM in the kernel, instead of using a userland
>> command?

It is rather dangerous unless you have a separate kernel w/o LVM or 
a boot option (e.g., -lm) that allows booting with LVM turned off.
Otherwise you can easily end up unable to reboot with a corrupt LVM
system. 

Re: [linux-lvm] Why vgscan not in kernel?

[Dirk Heinrichs]

Steven Lembark wrote:
> 
> >> upon boot, SCSI and IDE drivers check harddrives attached to the device
> >> for partitions. Wouldn't it be better to have similar functionality
> >> (scanning for VG's) for LVM in the kernel, instead of using a userland
> >> command?
> 
> It is rather dangerous unless you have a separate kernel w/o LVM or
> a boot option (e.g., -lm) that allows booting with LVM turned off.
> Otherwise you can easily end up unable to reboot with a corrupt LVM
> system.
I don't get the point here. Why is it dangerous to move the VG scan from
vgscan into the kernel module? Why should this corrupt LVM? Can you
explain this in more detail?

Bye...

	Dirk
-- 
Dirk Heinrichs		| Tel:	+49 (0)241 413 260
Configuration Manager	| Fax:	+49 (0)241 413 2640
QIS Systemhaus GmbH	| Mail:	dheinrichs@qis-systemhaus.de
J�licher Str. 338b	| Web:	http://www.qis-systemhaus.de
D-52070 Aachen		| ICQ#:	110037733

Re: [linux-lvm] Why vgscan not in kernel?

[josv]

If the VG scan process is in the kernel, that means that it
will *always* be executed (unless there is a kernel command
line option to disable it).

The issue then is that the VGSCAN in the kernel should be
*completely* resistant against every possible data corruption
and should be able to recover from it gracefully. This is not
as easy as it looks.

With VG scan in a separate executable, it is easier to prevent
it from executing...

++Jos


And thus it came to pass that Dirk Heinrichs wrote:
(on Mon, Jun 25, 2001 at 02:52:00PM +0200 to be exact)

> Steven Lembark wrote:
> > 
> > >> upon boot, SCSI and IDE drivers check harddrives attached to the device
> > >> for partitions. Wouldn't it be better to have similar functionality
> > >> (scanning for VG's) for LVM in the kernel, instead of using a userland
> > >> command?
> > 
> > It is rather dangerous unless you have a separate kernel w/o LVM or
> > a boot option (e.g., -lm) that allows booting with LVM turned off.
> > Otherwise you can easily end up unable to reboot with a corrupt LVM
> > system.
> I don't get the point here. Why is it dangerous to move the VG scan from
> vgscan into the kernel module? Why should this corrupt LVM? Can you
> explain this in more detail?
> 
> Bye...
> 
> 	Dirk
> -- 
> Dirk Heinrichs		| Tel:	+49 (0)241 413 260
> Configuration Manager	| Fax:	+49 (0)241 413 2640
> QIS Systemhaus GmbH	| Mail:	dheinrichs@qis-systemhaus.de
> J�licher Str. 338b	| Web:	http://www.qis-systemhaus.de
> D-52070 Aachen		| ICQ#:	110037733
> _______________________________________________
> 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

-- 
With all the things you are losing,
You might as well resign yourself,
And try and make a change...

Re: [linux-lvm] Why vgscan not in kernel?

[Heinz J. Mauelshagen]

On Mon, Jun 25, 2001 at 02:54:52PM +0200, josv@osp.nl wrote:
> If the VG scan process is in the kernel, that means that it
> will *always* be executed (unless there is a kernel command
> line option to disable it).
> 
> The issue then is that the VGSCAN in the kernel should be
> *completely* resistant against every possible data corruption
> and should be able to recover from it gracefully. This is not
> as easy as it looks.

Jos,

that's true. We'll try to provide that kind of redundancy/resistancy in a
post 1.0 release by having

 - checksums on all VGDA structures
 - two copies of the VGDA and
 - transaction oriented updates to VGDA contents

This should give us enough power to recover from a lot of potential cases.

BTW: there's will be a flag to switch init time activation per VG on and off.

Regards,
Heinz    -- The LVM Guy --


> 
> With VG scan in a separate executable, it is easier to prevent
> it from executing...
> 
> ++Jos
> 
> 
> And thus it came to pass that Dirk Heinrichs wrote:
> (on Mon, Jun 25, 2001 at 02:52:00PM +0200 to be exact)
> 
> > Steven Lembark wrote:
> > > 
> > > >> upon boot, SCSI and IDE drivers check harddrives attached to the device
> > > >> for partitions. Wouldn't it be better to have similar functionality
> > > >> (scanning for VG's) for LVM in the kernel, instead of using a userland
> > > >> command?
> > > 
> > > It is rather dangerous unless you have a separate kernel w/o LVM or
> > > a boot option (e.g., -lm) that allows booting with LVM turned off.
> > > Otherwise you can easily end up unable to reboot with a corrupt LVM
> > > system.
> > I don't get the point here. Why is it dangerous to move the VG scan from
> > vgscan into the kernel module? Why should this corrupt LVM? Can you
> > explain this in more detail?
> > 
> > Bye...
> > 
> > 	Dirk
> > -- 
> > Dirk Heinrichs		| Tel:	+49 (0)241 413 260
> > Configuration Manager	| Fax:	+49 (0)241 413 2640
> > QIS Systemhaus GmbH	| Mail:	dheinrichs@qis-systemhaus.de
> > J�licher Str. 338b	| Web:	http://www.qis-systemhaus.de
> > D-52070 Aachen		| ICQ#:	110037733
> > _______________________________________________
> > 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
> 
> -- 
> With all the things you are losing,
> You might as well resign yourself,
> And try and make a change...
> _______________________________________________
> 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

*** 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] Why vgscan not in kernel?

[Paul Jarc]

Dirk Heinrichs <heinrichs@qis-systemhaus.de> writes:
> Wouldn't it be better to have similar functionality (scanning for
> VG's) for LVM in the kernel, instead of using a userland command?

It would also be nice if VGs were initially active - then we wouldn't
need an initrd to have LVM on root... right?


paul

Re: [linux-lvm] Why vgscan not in kernel?

[Dirk Heinrichs]

Paul Jarc wrote:
> 
> Dirk Heinrichs <heinrichs@qis-systemhaus.de> writes:
> > Wouldn't it be better to have similar functionality (scanning for
> > VG's) for LVM in the kernel, instead of using a userland command?
> 
> It would also be nice if VGs were initially active - then we wouldn't
> need an initrd to have LVM on root... right?
This was my intention :-)

Bye...

	Dirk
-- 
Dirk Heinrichs		| Tel:	+49 (0)241 413 260
Configuration Manager	| Fax:	+49 (0)241 413 2640
QIS Systemhaus GmbH	| Mail:	dheinrichs@qis-systemhaus.de
J�licher Str. 338b	| Web:	http://www.qis-systemhaus.de
D-52070 Aachen		| ICQ#:	110037733