Disk Enclosures
cancel
Showing results for 
Search instead for 
Did you mean: 

RAID 5 Algorithm utilised in SmartArray devices

SOLVED
Go to solution
Litch
Occasional Advisor

RAID 5 Algorithm utilised in SmartArray devices

I'm moving away from the SmartArray devices I was using (1 died, 1 crashes every few weeks and the third was working fine but overall just isn't the best option for an Ubuntu OS host).

 

Anyway, I'm writing a recovery driver to assemble the array as is (using a JBOD HBA) so I can mount it and move over to another array type.

 

I can manually read in the first stripe of the old 4-disk RAID5 array and it's obvious that disk 1 is the start of the stripe, and disk 4 is parity. But that doesn't help as to if it's symmetrical or asymmetrical or a completely customised algorithm.

 

Is there a guru out there that knows what these P410's use? (v3.x firmware)

2 REPLIES
Litch
Occasional Advisor
Solution

Re: RAID 5 Algorithm utilised in SmartArray devices

Well, due to the overwhelming assistance I've received here, I've had to do things my own way. Which means hex dump each stripe and compare known file names and/or stumble across some text-based contents (thank god for Grep).

 

Somewhere over the first 51MB of data I've come up with 9 confirmed parity chunks and a handful of speculated ones.

 

128KB chunks, 384KB stripe, left-(a)symmetrical, delayed parity of 16.

 

That means for my 4-disk array:

 

D01,D02,D03,P01

D04,D05,D06,P02

[cut]

D46,D47,D48,P16

D49,D50,P17,D51

D52,D53,P18,D52

[cut]

D94,D95,P32,D96

D97,P33,D98,D99

[etc...]

 

Scary. I'm still not sure about the symmetry but that's as easy as reading a large text file from start to finish across multiple chunks (haven't got around to doing that yet). One speculates if HP did this deliberately to make it near bloody impossible to migrate using existing software RAID drivers.