[Linux-HA] Testing behaviour to disk faults

[Guy Coates]

Eduardo Grosclaude wrote:
> Hello,
> I intend to test my heartbeat+DRBD cluster and I'm curious about what
> will happen when the disk breaks. How should I monitor disk
> malfunctioning? How much of a test can I get? Is there some
> (non-destructive :) method to bring offline a directly-attached disk,
> or otherwise simulate a disk failure? IIRC software RAID has a special
> mode to allow for fault injections. I would like something alike for
> regular disk devices.
> Thank you very much
> 
I think the software-raid (MD) layer has a "faulty" target, which you can
configure to generate fake IO errors. So you could build your drdb device on top
of a single disk "faulty" software raid device...

Guy

-- 
Dr. Guy Coates,  Informatics System Group
The Wellcome Trust Sanger Institute, Hinxton, Cambridge, CB10 1HH, UK
Tel: +44 (0)1223 834244 x 6925
Fax: +44 (0)1223 496802