[dm-devel] Stopped disk causes LVM to go crazy

[Bryn M. Reeves]

Alireza Nematollahi wrote:
> hi guys,
> 
> I have a system with two SATA disks and I have LVM on the disks. I
> use sdparm (sdparm -C stop /dev/sdb) to spin-down one of them. It
> works fine and I can hear the disk spinning down and all...But when
> I run any LVM command, say lvscan or whatever, I get lots of error
> messages:

What did you expect? :)

If the volume group is active when you remove/spin down the disks LVM 
has no way of knowing that you wanted to do that. To the software, it 
just looks like a drive has failed or gone away while it was in use.

<snip>

> /dev/sdb2: read failed after 0 of 4096 at 2993881088: Input/output
> error /dev/sdb2: read failed after 0 of 4096 at 0: Input/output
> error /dev/sdb4: read failed after 0 of 1024 at 144082075648:
> Input/output error /dev/sdb4: read failed after 0 of 2048 at 0:

> Input/output error Even when I physically remove the disk it still
> happens. Is there any way I can stop LVM from querying both disks
> or at least to stop displaying the error messages?

Tell LVM to stop using the disk first. See the man page for vgchange 
for full details but you want to run:

vgchange -an <volume group>

To deactivate the volume group *before* you start removing devices 
that it is using.

Regards,
Bryn.