[linux-lvm] Bad disk?

[Mauricio Tavares]

On 11/11/2010 11:46 AM, Stuart D. Gathman wrote:
> On Wed, 10 Nov 2010, Mauricio Tavares wrote:
>
>>> Tell us exactly what you mean by "put a LVM on it".  Did you run
>>> pvcreate?  vgcreate?  lvcreate? You might find the output of "pvs"
>>> enlightening.  That will tell us what PVs you have created.
>>> And list /dev/mapper so we know what dm-0 is, and include the output of
>>> "lvs".
>>>
>> 	Let me put this way, I thought I did. I mean, after creating the
>> partition, setting it to LVM (8e), then running
>>
>> pvcreate /dev/sdc1
>> vgcreate export /dev/sdc1
>> lvcreate -L 400G --name vms export
>>
>> I used mkfs.ext4 to create partition (on /dev/mapper/export-vms) and off I
>> went. Do you think I missed a step?
>
> Great.  Now include output of "lvs"

raub at strangepork:~$ sudo lvs export
   /dev/dm-0: read failed after 0 of 4096 at 0: Input/output error
   LV   VG     Attr   LSize   Origin Snap%  Move Log Copy%  Convert
   vms  export -wi-a- 400.00g
raub at strangepork:~$

> BTW, if you really suspect a disk error, test for it directly.
> E.g., you can run
>
> # dd if=/dev/sdc1 of=/dev/null bs=256k
>
> to read through the partition or
>
> # smartctl -t long /dev/sdc
>
> To initiate a long self test of the disk (need smartmontools installed).
>
> A brand new disk that flunks self test is indeed defective.
>
> However, for real physical I/O errors, there would be errors logged
> in /var/log/messages referencing sdc (as opposed to dm-0), so I still
> think it is a logical error.
>