Help with RAID 0 Array

[Zack Hehmann]

Hello all,

I'm trying to get some data off an HPML310 G5. I was using it for testing at home. I had ESXI 5.5 installed on a flash drive and had the array as a datastore. I had configured a RAID 0 array with (3) 300GB 15K drives.

I had one of the drives fall out of the array due to bad blocks or something. I went ahead and ran a check disk or equivalent on the raid controller. The scan fixed whatever issues the disk had, but I was not able to add it back to the array. I would like to get my data off and moved to a different box. I would like to do it the quickest way possible.

I have another ESXI box. It's running 6.0 and I have (8) 2TB drives added to a NAPP-it vm. Can I just copy files from one datastore to another? Will the file permissions be an issue?

Here is a link to some pictures to provide more details.

Can I just delete the failed array, and create a new one but don't initialise..?

 

[pricklypunter]

You should be able to select the disk and add it back to the set, although the data may be damaged in such a way that it is no longer usable. I think at this point, and only if the data is critical, I would image all the disks in the array and use something like ReclaiMe to rebuild the array from the images and recover the data that way. Once the data is recovered, I would scrub the array and start over clean. However, it sounds like your disk is already on the way out. You can copy over the VM's and bring them back into your inventory, but again that will obviously depend on a successful recovery of said data. Personally I would rather just use single disks or a mirror of striped disks for a datastore, it's much easier to fix, should the unfortunate happen. Do you not have a back-up to restore from? That would obviously be your easy option

 

[Zack Hehmann]

Thanks for your response.

No, I do not have a backup. I used this environment for test and was ok losing everything. That's also why I went with RAID 0 because I needed the space and speed. There are a few things that I would like to have, but nothing important. I'm up for the challenge and excited to learn something new.

I like the imaging idea. I'm guessing the best way would be to use dd in Linux? I could turn each disk into an image file I can mount later, but I would probably just dd straight to a disk.

I came across this Arch Linux page on dd. The page also links to this and this. The Arch page talks about block size and mentions a warning on picking a certain block size will influence how read errors are handled. With that being said, it seems that I should set the IBS to 512b to decrease the issues that might occur from read errors. I'm guessing that the parameters on dd that allow you to specify the input and output block sizes can be whatever size you want, and it doesn't have to be the actual block size of the disk.

After reading further my best option is probably to use ddrescue instead of dd.

The drives I have to image are SAS. The only way I can attach these drives to a computer is with that RAID card. I plan to image them one at a time. Will the RAID card get in the way of imaging and will Linux be able to see the disk directly? I also have a HighPoint RocketRAID 2720 card I could use. Will that work in JBOD mode? I could also use my Dell H310 I flashed to IT mode, but I'm currently using it and don't want to pull it out if I don't have to.

After looking at the details from RAID Recovery Software - the only FREE tool I will need another drive to write the data to.

Have you had to do any of these things before?

 

[pricklypunter]

Well, at least this happened with something that's not critical, so if it proves to be totally borked, you have lost nothing but a few hours learning how to recover your data in the same situation later, there will be a later...

It's best to have direct access to the individual disks to image them, so use your HBA (H310) to achieve that. Otherwise there's a chance that your p410, or whatever card you're using in the HP, will see the failed array and attempt to do something with it, further destroying your data. Once you have the image files you can then mount them like you would the disk proper and use some recovery software to re-assemble the array, so that you can copy out whatever files you need from it. It's always safer to work from the images rather than the disks. If all goes wrong you can re-image and try again and again until you either get it right or give up, whichever comes first! The absolutely vital, critical thing to remember when recovering any data from a failed disk/ filesystem, is to never do anything that will risk writing data back to the disk you are recovering from.

Yes, I have had my moments with failed disks/ arrays and a gazillion filesystem cock-up's. I don't think anyone that's seriously involved with hardware that hasn't encountered this situation at some point or other. All I will say is, when it happens, and it will, is resist the urge to panic and do something carelessly that further corrupts your data, or it may just end in tears

Good luck with it...

 

[Zack Hehmann]

I ordered a H310 off of ebay and flased it to IT firmware. I ended up imaging all (3) 300GB drived using gddrescue. None of the 3 drives I imaged reported any errors during the imaging process. I made images of the first 2 drives straight to each 3TB drive. The last one that was having issues, I imaged to a file instead, that way I could write the data to different smaller drives if I had to. It would be more work to image the 3TB drive just to get the data off of it. I wrote the image file to a the last 3TB drive.

I installed all (3) 3TB drives in a Windows 10 pc with ReclaiMe installed. I added all 3 drives to the scan and let it do it's thing. After the scan was complete, it was not able to find anything.

I noticed something funny in disk management though. When I look at all 3 Drives this is what I get.

After looking at things, I determined that is how the first disk in the array looks on the original disk.

I also put the original disks in the Windows 10 box and ran the software. This is what I clicked on.

After it scanned everything, this was the result.

I want not able to get the SAS drives to show up in crystaldiskinfo. I did however get info from HD Sentinal.
It listed the drive as failing.

I understand that drives and the RAID card have array information. If the card is showing a failed array, couldn't I just delete the config on the card, and recreate the array, but don't initialise it?

Anyone have info on this? Also where could I find a manual for this card?

 

[msg7086]

The hard way:

Figure out the stripe size first, and write a program to interleave disk data manually. Disk images may be easier to work with than physical disks in this case.

 

[Zack Hehmann]

The hard way:

Figure out the stripe size first, and write a program to interleave disk data manually. Disk images may be easier to work with than physical disks in this case.

Sounds like fun!

I have images of all 3 drives. I have been doing some googling on the topic. I stumbled across this.