On 2011-01-27 06.02, Ted Unangst wrote:
> On Wed, Jan 26, 2011 at 10:00 PM, Amit Kulkarni <amitk...@gmail.com> wrote:
>> pardon my ignorance but if you restored your data already, why bother
>> investigating disk failure?

> Unless they are all the same person, there seems to be a sudden rash
> of people who want to bring a disk back from the dead because they are
> unwilling or unable to do the math on how much disks cost, how much
> time costs, and what the future integrity of their data is worth.  I
> don't know why this is, but I do know "disks die, buy new ones" is the
> correct answer to give them.

I fully understand the OP:s need to investigate this problem further,
regardless of whether there was any significant data loss or not.

It's a matter of uptime.

The indicated behaviour, that the system more or less freezes when
encountering a simple sector read error is indeed disturbing. For
example, my own reasons for using mirroring are exclusively so that a
system can remain online and operational in case of a disk failure.

If a disk in a mirror or redundant stripe set fails in a hotpluggable
hardware environment there really should be no need for service
interruption. The disk should be able to be replaced on the fly, or at
the very least during a controlled service window. In this case, that
obviously wouldn't work.

(The reason I'm butting in to this thread is that I'm currently
investigating a similar but probably totally unrelated problem, where a
system under high load (disk activity) claims there are sector read
errors, and then stops responding in a similar fashion to the OP:s
system. Saturate one, two or three disks with reads - no problem. Add a
fourth disk and after a while the problem appears. If I can determine
beyond reasonable doubt that this isn't a hardware problem, I'll submit
a bug report.)


Regards,
/Benny

-- 
internetlabbet.se     / work:   +46 8 551 124 80      / "Words must
Benny Lofgren        /  mobile: +46 70 718 11 90     /   be weighed,
                    /   fax:    +46 8 551 124 89    /    not counted."
                   /    email:  benny -at- internetlabbet.se

Reply via email to