Community
cancel
Showing results for 
Search instead for 
Did you mean: 
WSmit10
Beginner
1,772 Views

Clone all four RAID5 drives

Hi everyone,

I have a five year old SuperMicro server with Intel RSTe. It has been running Centos on a RAID5 (4x 2tb) bootable drive set for over a year.

Lately, I'm starting to see performance hits, and it looks like two of the drives have seek errors and command failures, so I bought four 2tb SSDs to do the replacement.

I have a second server, also with four bays, and I propose to put two old drives and two new SSDs in each one, Boot Live-CDs, and use dd or ddrescue to copy all of the data from the old set to the new set.

That actually seemed to work, because 100% of the data was recoverable, and it all fit in the SSDs. Sadly, the new set was not bootable.

This time, I used the BIOS I menu to construct a new RAID5 set with the new drives, using the same stripe size as the old set. I've restarted the same ddrescue, and I'll know in several hours whether it will work or not.

 

The question is: am I missing a critical step?

Thanks,

Will Smith

Tags (1)
0 Kudos
5 Replies
idata
Community Manager
34 Views

Hello Taiowa.

 

 

Thank you for contacting Intel Technical Support.

 

 

Based on the information provided by you, your system is in the middle of the process of cloning your RAID 5 after creating the RAID configuration on your RSTe beforehand, this seems to be a good idea and the only recommendation that we can provide you is to let the process finish and to check your RAID 5 virtual drive flags using a tool like https://gparted.org/display-doc.php%3Fname%3Dhelp-manual# gparted-manage-partition-flags Gparted. This in order to verify if the virtual drive that contains your operating system (OS) is indicating that is a bootable media.

 

 

As a friendly reminder, the support for your SuperMicro server should be provided by your system manufacturer.

 

 

Thank you for your patience and understanding.

 

 

Best regards,

 

 

Josh B.

 

Intel Customer Support

 

WSmit10
Beginner
34 Views

John,

Thanks for taking the time to reply.

 

The second result (create an array using BIOS I menu, use dd to duplicate 4 drives) also did not work.

It's interesting that individually copied / cloned drives won't boot. The BIOS doesn't recognize them as members of any array...suggesting that the metadata is in the wrong place(s).

Later, the faulty drive got so bad that it had to be removed. A similar spinning drive was installed. It was part of another array in a different server, so it automagically became part of a separate array. It was removed from that array, then added to the faulty array, and the automatic rebuild was successful.

The question of how to migrate the array to SSDs remains an open question. I guess one could remove the spinning drives one by one, adding the SSD each time, and allowing the array to rebuild.

It just seems like there should be a better way.

Thanks,

 

Will Smith

p.s. I opened a case with supermicro, but they said I should cruise the web...

idata
Community Manager
34 Views

Hello Taiowa.

 

 

Thank you for your reply.

 

 

Based on the information you provided us:

 

 

"The second result (create an array using BIOS I menu, use dd to duplicate 4 drives) also did not work.

 

It's interesting that individually copied/cloned drives won't boot. The BIOS doesn't recognize them as members of an array...suggesting that the metadata is in the wrong place(s)."

 

 

Actually this is an expected behavior if the drive that you are cloning the information to is not the same model, manufacturer or capacity of the original, since the metadata sometimes includes a unique ID for every drive on the array(this can be based on a UUID (Universal Unique Identifier), serial number, part number or assigned automatically by the tool used to create the RAID array) and is distributed through the metadata on each of the drives of the array in order to be able to handle details such as number of disks, stripe size and disk strip order and given those details any logical block address passed to it by the operating system can be calculated and mapped to "disk 2, location Y" or so on for example.

 

 

" I guess one could remove the spinning drives one by one, adding the SSD each time, and allowing the array to rebuild."

 

 

This is a slower but safer way to do it and an excellent option in your case that the original cloning idea using DDrescue failed.

 

 

"It just seems like there should be a better way."

 

 

There are some professional software options on the market that are able to do this process, but we are not able to recommend a specific one.

 

 

We hope you find this information useful.

 

 

Thank you for your patience and understanding.

 

 

Best regards,

 

 

Josh B.

 

Intel Customer Support

 

idata
Community Manager
34 Views

Hello Taiowa.

 

 

Thank you for having contacted Intel Technical Support.

 

 

We have not heard from you since our last communication and we would like to know if you need further assistance or if we can close this case?

 

 

Important note: Should further assistance or clarification be required, we will greatly appreciate if you reply to this post instead of writing a new one unless your inquiry is completely unrelated. This way we will prevent generating a duplicate post and we will not lose the train of thought.

 

 

We will be looking forward to your reply.

 

 

Best regards,

 

 

Josh B.

 

Intel Customer Support.
WSmit10
Beginner
34 Views

Hi John,

Turns out that I missed your response above - thanks for rattling my cage.

For those playing along at home: ddrescue is functionally identical to the common dd command except that it is intended to recover difficult-to-read data from spinning drives.

Thanks, John, for confirming my suspicion that the medadata prevents a clone from a different vendor / model, etc from being able to function. That was fairly obvious immediately, but I was hoping you could point me at the metadata and I would be able to unscramble it. No such luck.

Since the original array is repaired now, and seems robust for now, and because I have a duplicate server, I think I will do a fresh install of Centos7, and copy all the goodies for KVM and the VMs via the network. Then with two servers, I can have twice the fun.

Again, Thanks for the nudge - yes, please close the case.

 

Regards,

Will