[Vdo-devel] a couple of dmesg errors: 'error after readData'

[Łukasz Michalski]

On 16/09/2019 22.22, Sweet Tea Dorminy wrote:
> Greetings Łukasz;
> 
> I'm sorry for the issue with your array. I hope its rebuild is going well.
> 
> Those log entries indicate that VDO received several reads, for blocks 342360164, 342360165, 342360167, 342360168; when VDO attempted to read data off storage for these locations, the storage returned an IO error (-5). VDO goes into read-only mode if  a metadata read or write, or (in async mode) a data write, is failed by the storage. But for data reads, VDO just passes the error up to the filesystem or other user. It's not a VDO metadata inconsistency, just a problem reading from the underlying storage, and no maintenance on VDO need be done.
> 
> Let us know if you have more questions!
> 

Many thanks for a quick answer!

I do not see any errors from mdadm or sata device reported by kernel.
All of those IO errors (-5) magically disappeared after mdadm raid5 recovery ended.

After that I ran dd on /dev/mapper/vdo_device to read part of disk:

dd if=/dev/mapper/vdo_device of=/dev/null bs=4096

No errors at all.

Next 8h backup was run tonight and still no errors.

Could it be that this IO error is a timeout caused by mdadm rebuild?

Regards,
Łukasz