Mailinglist Archive: opensuse (1196 mails)

[Per Jessen <per@xxxxxxxxxxxx>]

David C. Rankin wrote:

> Simply put, if small disk errors go uncorrected in the dmraid array
> and are allowed to remain uncorrected and multiply, then dmraid has a
> gaping hole in its robustness making it look far inferior to software
> raid.

Surely you can still use SMART to monitor the drives, right?

> Does anybody know more or have any links to detailed information on
> how bad blocks are handled in dmraid??

It's done automagically by the IDE drive. 

> All of this discussion is prompted by 2 recent supposed disk
> 'failures' in dmraid setups where one disk will be pristine, but the
> other will have hundreds of simply corrected errors that have never
> been fixed over the life of the array, that after disabling the array
> and running e2fsck are fully correctable. 

If I understand this right, you have a two-disk RAID1 array run by a
hardware/on-board controller that you access via dmraid.  Unless
something is wrong with the mirroring, I don't see how one disk can
have file-system errors that the other does not. 

> In all the dmraid 
> documentation I have read, I have never seen a warning or note that
> says:
>
> "Periodic disabling of the dmraid array will be required to allow fsck
> or other disk maintenance processes to be run on the individual disks
> of the array because enabling dmraid in the BIOS prevents disk
> maintenance on individual disks in the array."
>
> I'm concerned that something like that is needed. The question is, "Is
> that true?"

Unless your RAID setup also prevents SMART monitoring, no. 


/Per

-- 
Per Jessen, Zürich (-0.8°C)

--
To unsubscribe, e-mail: opensuse+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: opensuse+help@xxxxxxxxxxxx