Re: getting I/O errors in super_written()...any ideas what would cause this?

[Chris Friesen <chris.friesen@genband.com>]

On 12/03/2012 02:22 PM, Ric Wheeler wrote:
> On 11/28/2012 03:29 PM, Chris Friesen wrote:
>> On 11/28/2012 02:27 PM, Mathias Burén wrote:
>>
>>> The drives look healthy, but am I reading that right? More than 10
>>> self tests per hour?
>>
>> Yeah....we cranked it up to try and increase how frequently we see the
>> problem.
>>
>> From what I understand normally it runs once a day.
>>
>> Chris
>
> Did the vendor suggest to you that running a self test on an active
> drive would be OK? I would expect errors in this case - specifically
> time outs....

I'm not the main developer in that area, but from what I understand the 
code has been like this for ages.  (It's entirely possible we've been 
lucky up till now since we support limited hardware types.)

The fact that you'd expect time outs is interesting--is that from the 
delay switching from doing the self-test to doing the actual request?

Is the expectation that the OS should not be sending any other commands 
to the disk while doing the self-test?

I was recently looking at the SCSI spec trying to learn a bit about this 
issue and the section on background self-test (spc-4, section 5.15.4.3) 
seems to indicate that a READ or WRITE command should cause the 
background self-test to be aborted and the command to be processed 
within 2 seconds.  In our case it doesn't seem to be aborting (at least 
it shows as "Completed" in smartctl)--is this expected?

Thanks,
Chris
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html