Cloning 1 hard disk in a software RAID

[Liam Proven]

2009/7/29 Darren Upton <dbupton at googlemail.com>:
> Liam Proven wrote:
>> 2009/7/28 Colin Law <clanlaw at googlemail.com>:
>>> 2009/7/28 Liam Proven <lproven at gmail.com>:
>>>> 2009/7/28 Colin Law <clanlaw at googlemail.com>:
>>>>> 2009/7/28 Liam Proven <lproven at gmail.com>:
>>>>>> 2009/7/28 Alvin Chang <alvin.chang at gmail.com>:
>>>>>>> On Tue, Jul 28, 2009 at 16:20, Liam Proven <lproven at gmail.com> wrote:
>>>>>>>> I have a degraded & rather unstable software RAID on my server. 1 disk
>>>>>>>> does not register at all any more & the md subsystem won't mount it as
>>>>>>>> being "stale". Another sometimes mounts but won't stay up for long.
>>>>>>> What does cat /proc/mdstat say? Which RAID level is it?
>>>>>>>
>>>>>>>> <snip>
>>>>>>>> dd if=/dev/sdd of=/dev/sde
>>>>>>>>
>>>>>>>> to copy the whole of the unstable drive onto a new one.
>>>>>>>>
>>>>>>>> Should this work, or is it a disastrously bad idea or something?
>>>>>>> Well, that really depends... if the source drive is broken, you might need
>>>>>>> to use something other than dd to clone the disk so that it ignores bad
>>>>>>> sectors, e.g. Gohst 4 Linux...
>>>>>> I don't think I have any bad blocks, it just won't always spin up.
>>>>>>
>>>>>>> In fact, if you are using RAID5 or above... simply re-initialize the array
>>>>>>> with a working blank drive should do the trick.
>>>>>> Tried that; nothing seems to happen & this is the only copy of the
>>>>>> data that I have so I am wary of experimenting too much...
>>>>> I am sure will already have done this but backup the _really_
>>>>> important stuff onto CD or another machine or something before doing
>>>>> anything else
>>>> I can't, the array won't stay up long enough... Thus the last-ditch
>>>> efforts to try to get some stuff off the blasted thing.
>>>>
>>> I don't know much about RAID so this may be a silly suggestion.  Could
>>> you take the sometimes working disk out and plug it in to a different
>>> PC (or the same one) as a single disk rather than part of the RAID
>>> set?  Or can it not work on its own?  Not that this will help if the
>>> disk won't keep going.
>>
>> They are 4 ordinary 40GB EIDE drives. I've already moved the whole
>> array to a new server, a Dell PowerEdge 600SC. I was planning to
>> attach the iffy one(s) in isolation to clone them, yes.
>>
>>
> I've been watching this thread - I don't know too much about software
> RAID, but run 11 servers at work using hardware RAID.
>
> I don't this you've mentioned, although you've been asked which level of
> RAID you're using (apologies if I missed it).  With 4 drives, I am
> assuming level 5.

Oops, sorry, yes, it's RAID5.

> If this happened on one of my RAID systems, it would indicate that two
> drives had gone rather than just the one.

That is correct. One has dropped out apparently permanently & another
tends to fall over after a period of operation - perhaps an hour or
so, sometimes.

>  RAID 5 can cope with losing a
> drive for a little while - simply replace the drive ASAP and rebuild the
> array.

Alas I got 2 at once. And the backup failed.

I hate modern EIDE hard disks. I seem to see more failing every year,
although the late 1990s and early 2000s were the worst. It's why I was
using RAID at all. I should have gone for RAID6.

> Can you send the contents of /etc/raidtab and /proc/mdstat.  Also is
> there any output in dmesg?

I will the next time I bring the server up.

-- 
Liam Proven • Profile: http://www.linkedin.com/in/liamproven
Email: lproven at cix.co.uk • GMail/GoogleTalk/Orkut: lproven at gmail.com
Tel: +44 20-8685-0498 • Cell: +44 7939-087884 • Fax: + 44 870-9151419
AOL/AIM/iChat/Yahoo/Skype: liamproven • LiveJournal/Twitter: lproven
MSN: lproven at hotmail.com • ICQ: 73187508