Determining if a stripe/RAID0 has failed

Determining if a stripe/RAID0 has failed

[Curtis]

Hi All,

I'm wondering what the best way to determine when a RAID0 has failed?

We have some stateless servers that use a stripe/RAID0, but we'll need
to know if it failed so we can pull it out of the "cluster" and
rebuild it. It would be better to find out sooner than later that the
stripe has failed.

I know from reading the man page that I can't use mdadm to monitor the
stripe. Is it basically just that the device becomes unusable in some
fashion?

Thanks for any pointers,
Curtis.

Re: Determining if a stripe/RAID0 has failed

[NeilBrown]

[-- Attachment #1: Type: text/plain, Size: 1064 bytes --]

On Tue, 9 Jul 2013 15:33:29 -0600 Curtis <serverascode@gmail.com> wrote:

> Hi All,
> 
> I'm wondering what the best way to determine when a RAID0 has failed?
> 
> We have some stateless servers that use a stripe/RAID0, but we'll need
> to know if it failed so we can pull it out of the "cluster" and
> rebuild it. It would be better to find out sooner than later that the
> stripe has failed.
> 
> I know from reading the man page that I can't use mdadm to monitor the
> stripe. Is it basically just that the device becomes unusable in some
> fashion?
> 

How would you determine if a lone drive had failed?
Presumably by error messages in the kernel logs, or similar.
Use exactly the same mechanism to test if a RAID0 has failed.

(A "RAID0" doesn't fail as whole.  Bits of it might, other bits might keep
working, just like a drive which can lose some sectors but other sectors keep
working.  Certainly a whole drive can fail if it's logic-board dies.
Similarly a whole RAID0 can fail if the SATA/SCSI/USB controller dies.)

NeilBrown

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 828 bytes --]

Re: Determining if a stripe/RAID0 has failed

[Curtis]

On Tue, Jul 9, 2013 at 3:46 PM, NeilBrown <neilb@suse.de> wrote:
> On Tue, 9 Jul 2013 15:33:29 -0600 Curtis <serverascode@gmail.com> wrote:
>
>> Hi All,
>>
>> I'm wondering what the best way to determine when a RAID0 has failed?
>>
>> We have some stateless servers that use a stripe/RAID0, but we'll need
>> to know if it failed so we can pull it out of the "cluster" and
>> rebuild it. It would be better to find out sooner than later that the
>> stripe has failed.
>>
>> I know from reading the man page that I can't use mdadm to monitor the
>> stripe. Is it basically just that the device becomes unusable in some
>> fashion?
>>
>
> How would you determine if a lone drive had failed?
> Presumably by error messages in the kernel logs, or similar.
> Use exactly the same mechanism to test if a RAID0 has failed.

Ok, that makes total sense, thanks. :)

>
> (A "RAID0" doesn't fail as whole.  Bits of it might, other bits might keep
> working, just like a drive which can lose some sectors but other sectors keep
> working.  Certainly a whole drive can fail if it's logic-board dies.
> Similarly a whole RAID0 can fail if the SATA/SCSI/USB controller dies.)

Noted.

Thanks again,
Curtis.

>
> NeilBrown



--
Twitter: @serverascode
Blog: serverascode.com