[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [Nbd] turn off debug messages?



On 11/11/06, Wouter Verhelst <wouter@...3...> wrote:
On Fri, Nov 10, 2006 at 10:33:22PM +0000, John Hannfield wrote:
> Hello,
> I have a nbd enabled kernel, 2.6.16.29 running on debian-amd64
> Etch with
>
> nbd-client                      2.8.7-1
> nbd-server                      2.8.7-1
>
> In /dev/   I have nbd[1-15]
> However on boot up or when doing any LVM commands,
> pvscan, lvscan etc, I get lots of nbd errors.
>
>   /dev/nbd0: read failed after 0 of 4096 at 0: Input/output error
>   /dev/nbd0: read failed after 0 of 4096 at 2199022141440: Input/output error
>   /dev/nbd0: read failed after 0 of 4096 at 0: Input/output error
>   /dev/nbd1: read failed after 0 of 4096 at 2199022141440: Input/output error
>   /dev/nbd1: read failed after 0 of 4096 at 0: Input/output error
> .....
>
> Is there a way to disable these error messages?
> Even if I remove the /dev entries, I still get the errors on boot.

That is related to LVM trying to access unconnected NBD devices, not to
NBD itself. There is a workaround; some googling on the combination of
NBD and LVM a while back gave me a link to the post on the LVM
mailinglist that explains it.

Actually it is the kernel's nbd driver that is exposing these nbd
devices (see /sys/block/nbd*) even though the device(s) haven't been
started.  You can limit the amount of nbd devices that get exposed
when you load the nbd kernel module by using the 'nbds_max' module
parameter.

But even when you do so LVM will still try to read LVM labels off all
known block devices (including your now limited number of unconnected
nbd devices).  So its actually the nbd driver that isn't doing the
right thing; it should only register an nbd device with linux (under
/sys/block et al) when the nbd driver gets an ioctl to start an nbd on
behalf of the userspace nbd-client daemon.

Mike



Reply to: