ML570G3/RAID 1+0 - which disks are mirroring and which are striping?

[bresroc]

To make a long story short, a server upgrade i was doing failed. the server was set up with a 4 disk raid 1+0. so i had the idea to take the primary set out of the bad server and swap them into a good server that still has the secondary set present and simply rebuild the bad primary set using the good secondary set.
Once the set was rebuilt, i could then pull them back out of the good server and move them back to the original server. I could then use this rebuilt primary set back in its original server to rebuild the original secondary set.
what I'm trying to verify is which disks are being striped and which is the set that mirrors that stripe. (I thought i knew, but now I'm not sure).
Also, how do i prevent it from trying to rebuild the wrong set (i.e. prevent erasing the good data and replacing it with the bad data, which is opposite of the intended goal)

Physically, the layout of the disks look like this when you are staring at the front of the server:
______________________________________________________
[slot 2, scuzi 0]
[slot 1, scuzi 0][slot 1, scuzi 1][slot 2, scuzi 1]
______________________________________________________

what I have been able to determine about the layout so far using start-smart and educated guessing is:

[slot 1, scuzi 0] + [slot 1, scuzi 1] = primary stripe set

[slot 2, scuzi 0]+[slot 2, scuzi 1] = secondary stripe set

[primary striped set] mirrored by [secondary striped set] for fault tolerance.
That is my assumption anyway, please let me know if i have that incorrect.
Any help would be greatly appreciated.
(I'll keep google-ing and searching in the meantime...)

Thanks
JD

 

[technome]

be real careful....
Unless you documented the mirrored pair sets, you are finished if you pull the incorrect drives. I would not guess.

Much safer, add a SATA or IDE drive, using cloning software, clone the array to the SATA/IDE drive, move it to the other server and reverse clone from the SATA/IDE to the raid array.

........................................
Chernobyl disaster..a must see pictorial

http://www.kiddofspeed.com/default.htm

 

[bresroc]

just a small correction, i made a mistake about the scuzi layout. It did not affect anything else though.
______________________________________________________
[port 1,scuzi:1]
[port 1,scuzi:0][port 2,scuzi:0][port 2,scuzi:1 ]
______________________________________________________

After doing some testing I've determined that
Port 1 drives can be run on their own, without the other two drives in port 2. So it appears that port 1, scuzi:0,1 are a striped set, and they are mirrored by the striped set on port 2 scuzi:0,1

Backing up to SATA or adding to the array is not an option.

The drives in question are originally from an ML570G1 server, that i put in an ML570G3 server to do the upgrade, and i am now attempting to transfer the "now upgraded" hard drives back into the original ML570G1.
I would not think the generation of the server would affect the way it configures a raid level's striping and mirroring, though. I was under the (possibly mistaken) assumption that it would always be the same.

One other question (if anyone reads this far)...If i can put in 2 of the 4 total drives, and cause the server to stop at boot and tell me that the two drives that i left disconnected "have failed, and do i want to enable the interim recovery mode"...
Is that what i want to do? Enable the "interim recovery mode", then reconnect the second set of drives to let them rebuild automatically?
I have gotten to this point, but i was afraid to actually boot since i didn't want to wipe out the array.

 

[blister911]

Boot off the smartstart CD. Go into the ACU (Array Configuration Utility). See if you have the option to split the mirror. If so, you should be able to select 1 of the pairs of drives, and their indicators should light up. That will tell you which pair is striped. The other non-lit drives are your mirror. Don't save the changes or you will actually split the mirror.


Light travels faster than sound. That's why some people appear bright until you hear them speak.