Help with Raid 5

[DocGreen]

Thanks for the insight guys. I'll plan for recovery at this point. My question now is, how much space will I need to for recovery?

The array is four 1TB drives in a RAID 5, with 2.795 TB of storage capacity... will I need a 3TB drive or 4TB to recover the data?

 

[coffee]

Thanks for the insight guys. I'll plan for recovery at this point. My question now is, how much space will I need to for recovery?

The array is four 1TB drives in a RAID 5, with 2.795 TB of storage capacity... will I need a 3TB drive or 4TB to recover the data?

Price wise its not a big difference. I would just get the 4 tb to be sure. The 3 tb of storage might not be enough after formatting.

 

[DocGreen]

Good point

 

[lcoughey]

You need 4 x 1TB drives for the clones and a destination large enough to hold the data...if it only has 1.5TB of data, a 2TB destination drive would be enough.

 

[coffee]

I think it would be better to just get the one large drive and get the damaged raid mounted and copy everything over. Buying a whole new raid set is kinda expensive when you just want to recover your information. Then take the existing raid drives and rebuild a new raid and copy the data back to the new raid.

Just figuring if you buy a whole new set of drives your going to wrap up about 300 / 400 bucks when one drive for about 120 bucks or so will hold your recovered data temporarily.

 

[Markverhyden]

If you want to rebuild and have the drives to use then going the 4 x 1TB as @lcoughey said. The problem is because the array is marked failed you will not be able to use "normal" procedures. The link I posted above covers trying to recover/rebuilt a software raid with mdadm. Includes clearing the failed status of the array.

Another option is to use DR software, like R-Studio. I've successfully used it in the past with software RAID to recover data. You just image each of the good drives. Then mount the images and it should see they are RAID'd, if not you can probably force it. Scan and recover data. Once you are confident that the data is good you can rebuild the array from scratch. If this data is very important then you will need to have 2 x 3tb. One for making the images and the other for the data recovery.

 

[lcoughey]

I think it would be better to just get the one large drive and get the damaged raid mounted and copy everything over. Buying a whole new raid set is kinda expensive when you just want to recover your information. Then take the existing raid drives and rebuild a new raid and copy the data back to the new raid.

Just figuring if you buy a whole new set of drives your going to wrap up about 300 / 400 bucks when one drive for about 120 bucks or so will hold your recovered data temporarily.

I get lots of work from techs who make this mistake...usually at their client's expense.

 

[coffee]

I get lots of work from techs who make this mistake...usually at their client's expense.

I guess I am not following you. Can you elaborate a bit?

 

[DocGreen]

You need 4 x 1TB drives for the clones and a destination large enough to hold the data...if it only has 1.5TB of data, a 2TB destination drive would be enough.

I understand what you and Mark are saying... what about this:

1. Image each 1TB drive using a single 4TB as the destination
2. Once all 4 images are saved to the 4TB, mount and recover via R-Studio (question... if the 1 bad drive won't image, can R-Studio rebuild using the other 3?)
3. Using the 3 good 1TB drives and the 1 replacement 1TB, create a new 4-disk Raid5
4. Use that new raid as the destination for data recovered in R-Studio.

That way I'm only buying one drive. Does this sound acceptable?

 

[coffee]

I understand what you and Mark are saying... what about this:

1. Image each 1TB drive using a single 4TB as the destination
2. Once all 4 images are saved to the 4TB, mount and recover via R-Studio (question... if the 1 bad drive won't image, can R-Studio rebuild using the other 3?)
3. Using the 3 good 1TB drives and the 1 replacement 1TB, create a new 4-disk Raid5
4. Use that new raid as the destination for data recovered in R-Studio.

That way I'm only buying one drive. Does this sound acceptable?

Ok, I misunderstood some things I guess.

I think if you make clones of the raid drives for safety sakes you can work with the clones to re assemble / recover the raid. Then you may not need the 4 tb drive. It all depends on how you are recovering the raid. Using something like mdadm in linux you would install the cloned raid drives or the original set with a good spare (doesnt matter) in a running linux system and use mdadm to rebuild / assemble the raid in degraded mode. But if your using something like R-Studio ( I have not ever used it) the procedure might be different and you would recover the raid and save the files to the 4 tb then rebuild the raid and immediately copy the files back to the raid from the 4tb.

mdadm will just use the clone drives and spare to rebuild the raid. Recovery programs might operate differently I dont know as I have never used anything else in my experience.

 

[kwisher]

Maybe I've missed something in this very long post but: If you are going to clone each drive in the failed array, which I highly suggest, you are going to need extra storage large enough to hold all images plus enough storage to hold the data recovered. Is one 4TB drive going to be enough? The individual drive images would not be able to be deleted until after the data recovery so, 4 x 1TB image = 4TB apprx. + xTB data. This only applies if you are going to try raid recovery using the images and not the drives themselves.

My suggestion would be to image all drives to spare storage for safe keeping in case of recovery failure. Boot system with the failed array with a live Linux distro of your choosing and install mdadm. Use mdadm to repair the failed array. When you can access your data, delete the saved drive images and copy data to spare storage. Reboot system and rebuild bad array using software raid and don't use the mobo raid utility.

 

[Markverhyden]

I understand what you and Mark are saying... what about this:

1. Image each 1TB drive using a single 4TB as the destination
2. Once all 4 images are saved to the 4TB, mount and recover via R-Studio (question... if the 1 bad drive won't image, can R-Studio rebuild using the other 3?)
3. Using the 3 good 1TB drives and the 1 replacement 1TB, create a new 4-disk Raid5
4. Use that new raid as the destination for data recovered in R-Studio.

That way I'm only buying one drive. Does this sound acceptable?

In my book you are short storage if you really value the data.

Say you go the data recovery method. Image each 1tb drive to the 4tb. Yes, R-Studio should be able to load the 3 drives as RAID to recover data if the drive headers are still intact. As that is how RAID5 is intended to work.

Where are you going to send the data to? If you take the 4 x 1tb and rebuild the array to copy it to, you've lost your backup and only have one copy of data. Which is risky. Personally I'd have another 3tb drive to put the data recovery on since the total store is a bit under 3tb. Another thing to pay attention to is the data being recovered. It's completely normal for R-Studio to recognize an amount of data that is many times the actual store since it will report that same files under different circumstances.

 

[DocGreen]

Reboot system and rebuild bad array using software raid and don't use the mobo raid utility.

I'm actually thinking about moving to a NAS for my main storage. Anyone have an opinion on these? https://www.evertek.com/viewpart.asp?auto=100005&cat=19 I can get a pair of them for about $200 from Evertek. (That would actually solve multiple problems, as WSE2012 doesn't allow anonymous shares)

Ok, I misunderstood some things I guess.

I think if you make clones of the raid drives for safety sakes you can work with the clones to re assemble / recover the raid. Then you may not need the 4 tb drive. It all depends on how you are recovering the raid. Using something like mdadm in linux you would install the cloned raid drives or the original set with a good spare (doesnt matter) in a running linux system and use mdadm to rebuild / assemble the raid in degraded mode. But if your using something like R-Studio ( I have not ever used it) the procedure might be different and you would recover the raid and save the files to the 4 tb then rebuild the raid and immediately copy the files back to the raid from the 4tb.

mdadm will just use the clone drives and spare to rebuild the raid. Recovery programs might operate differently I dont know as I have never used anything else in my experience.

I'm a little rusty on my Linux, but willing to give mdadm a shot... it looks like it's included w/ Parted Magic.

 

[fencepost]

I suspect the NAS recommendations are going to tend towards Synology or QNAP. I'll also note that Cloudberry (probably among others) actually runs on both of those, and lists the WD NAS units as "Coming Soon".

Those Verbatim units also appear to be basically nonexistent on the Verbatim site except for a download of the user manual when I searched, out of stock on Newegg with no reviews, and someone offering to sell them on Amazon for $600 each (also with no reviews). I saw one mention that they supported up to RAID 5 though I confess to not looking at the manual. If they only do 5 not 6 that's a bad sign with modern drives.

I see nothing about what they're running, updates, anything like that. I'd stay away, particularly if they're being clearanced out of Evertek for $50/each.

 

[DocGreen]

OK, I'm feeling a little silly...

How do I go about imaging these drives in a way that R-Studio can open them? (*.rdr, *.dsk, *.dat, *.bin, *.dfs, *.fss, *.hex, *.arc)
Normally I'd just use Clonezilla or Imagex, but Clonezilla errors out with "No Input Device!" (I get the same error regardless of which disk I try to image. I'm assuming I need to use a different program?

 

[fencepost]

Looking at the description of R-Studio, it looks like it can create the image files itself based on their bulleted list of capabilities

And they have links to a bunch of stuff at the lower left column, including a presentation on RAID recovery

 

[DocGreen]

Looking at the description of R-Studio, it looks like it can create the image files itself based on their bulleted list of capabilities

And they have links to a bunch of stuff at the lower left column, including a presentation on RAID recovery

Yeah, I can see how to create an image from within R-studio... but the problem there is when I loaded up with the raid drives in, R-studio shows them as a single 3tb volume (intel raid) instead of individual disks. That was the case whether I loaded up with one or all of the disks, by the way.

I've got drive 1 imaging w/ ddrescue now... but it's slow as hell, and it's outputting a *.img file. I'm not even sure if R-studio will be able to open it.

 

[coffee]

I've got drive 1 imaging w/ ddrescue now... but it's slow as hell, and it's outputting a *.img file. I'm not even sure if R-studio will be able to open it.

Curious,

What is your command line for ddrescue and what type of interface are you connecting the drives too?

My basic command line for dd is the following:

dd if=/dev/sd? of=/dev/sdX conv=noerror,sync bs=2M

I have 2 USB3 external doc stations but I prefer eSATA because you can check the smart on the drives too.

coffee

 

[DocGreen]

Curious,

What is your command line for ddrescue and what type of interface are you connecting the drives too?

My basic command line for dd is the following:

dd if=/dev/sd? of=/dev/sdX conv=noerror,sync bs=2M

I have 2 USB3 external doc stations but I prefer eSATA because you can check the smart on the drives too.

coffee

root@partedmagic: /media/sdf1# ddrescue -d -r3 /dev/sda raid5disk1.img raid5disk1.logfile

Took that straight from a TN guide to using ddrescue, fwiw

Drive being imaged is connected via SATA... destination is a 4TB external USB (2.0... that machine doesn't have 3, sadly)

 

[Markverhyden]

Yeah, I can see how to create an image from within R-studio... but the problem there is when I loaded up with the raid drives in, R-studio shows them as a single 3tb volume (intel raid) instead of individual disks. That was the case whether I loaded up with one or all of the disks, by the way.

I've got drive 1 imaging w/ ddrescue now... but it's slow as hell, and it's outputting a *.img file. I'm not even sure if R-studio will be able to open it.

That's good news. Means the disk headers are most likely intact if R-Studio seems it as an array. So, hopefully, when you mount them, which R-Studio can do, they will mount as an array. It's odd that it is showing it as a 3tb array.

Is this the link you were referring to? - https://www.technibble.com/guide-using-ddrescue-recover-data/