Major problem with IBM M1015- Configuration Not Importable

E

eli_singer

n00b
Joined
Jul 7, 2013
Messages
17
had X4 HGST 4TB drives running at RAID 10 on an Intel DZ77RE-75K.
i got this every time windows (7 ultimate X64) started:



after a while drives started to break from the array and needed to be rebuilt in LSI MegaRAID. very fast i got to a point where i had a failing drive while another one is being rebuilt.
eventually i ended up with 3 failed drives and only 1 working.

i pulled the array and the card out to fix it before i'd loose everything.

hooked it up on a different system, an old Gigabyte EP45-DS3R.
this is what i get in the WebBIOS:



i'm afraid to import since it might not be able to rebuild the array...

how should i proceed from here? how do i rebuild the array?

please help...

Eli
 
How many drives do i need to do that?
Backup means 2 drives and to recreate it i'm suppose to use the original drives?
I have 2 2tb drives and another 3 1gb external drives abd i'm trying to figure out if that's enough...

Eli
 
if ever disks break, and you want to recover the data, the first and most important thing to do is create a forensic-quality image. you don't repair arrays or repair disks. you get the data onto a working disk and then shelf the bad disk.

this needs to be sticky........... along with not using sibling drives in an array.

I wish posts like yours stopped happening. you have some work ahead of you. search for gnu ddrescue. it keeps track of what is copied and what is failed. if you need to image 4tb disks and you don't have empty 4tb disks to image to, you're in trouble.

another head added to the RAID guillotine pile, folks
 
well, i can't import the array so i don't know how should i backup the drives since i can't get into Windows with the array set up.

what do you mean not using sibling drives in an array?

it's a RAID 10 so its is suppose to look like a big 8tb drive anyways and since there aren't any 8tb drives yet the backup will have to be spitted to several drives anyways won't it?
so why does it matter if i'm using a few 2tb drives instead of 4tb?

i will try to get a 4tb drive if i must but it won't be easy.
but still, right now i don't know how to backup the drives since i an't get pass the WebBIOS and into Windows without configuring the array in some way and i don't want to configure it as a new array and harm my chances of recovery...
what are my options here?

Eli
 
You need to plug the drives into another controller(or another computer), usually it means buying a cheap sata port add-on card. Once you have the drives plugged in but not part of a RAID you can use the r-studio program I linked to try and recreate the array in software, from there you get your data off of it by copying it file by file to other drives.

Alternatively, if you really care about your data and it's irreplaceable, then you should do as bamtan2 says and instead of playing with your array then you should make a direct 1:1 copy of the disks before attempting to recreate the array with r-studio.

The difference is the amount of disks you need. First method you simply need a place to hold the data. second method you need both 4 identically sized disks and also a place to backup the data to.
 
The caution here is that you stand a very real chance of loosing all data on those drives.

These next few steps on what you do will be crucial.

Some people have good luck trying to recover on their own, but others who really care about what's on their drives will pay a professional recovery center at this point.

The time honored saying "RAID is not a backup" rings true.
 
If you care at all about your data do this:

instead of playing with your array then you should make a direct 1:1 copy of the disks before attempting to recreate the array
Click to expand...
 
i want to do it as safe as i can get, i just want to make sure i'm doing it right.

i have the X4 4tb drives that were in the RAID 10 array. meaning i have 8tb of data (actually it's less because they're not completely full but still, let's say it's 8tb).

i should just plug them out of the controller and in to another machine?
can i just plug them to the SATA ports on the MB?
how should i configure them in the BIOS? as RAID? as AHCI?

that will suppose to result in 4 drives recognized and accessible by Windows?

now i'm a bit confused.
going by the safe method, i'm suppose to make a 1:1 copy (with r-studio?) of the drives to some other drives? can i split this image to several 2tb drives?

and after that i should give r-studio a chance to rebuild the array on the original X4 4tb drives? am i getting this right?

thanks for all the help...

Eli
 
how should i configure them in the BIOS? as RAID? as AHCI?
Click to expand...

AHCI


going by the safe method, i'm suppose to make a 1:1 copy
Click to expand...

Linux ddrescue - easiest way will be a live usb stick. I recommend sysrescuecd

and after that i should give r-studio a chance to rebuild the array on the original X4 4tb drives?
Click to expand...

Original or copy. I would try the copy and not mess with the original. That would get you the best chance of success if you fail to recover and need to spend a few thousand on a recovery company.
 
You would need 4 identically sized 4TB disks to do this correctly then.

Yeah you could theoretically split up disk images with dd, but you will need the full 16TB of places to put the data still, and you will not be able to use these images unless you can re-image them onto 4 other 4TB drives, and even then you might not get it to work.

Cheapest most effect route would be just to buy 4 more 4 TB drives to image them.

And look at it this way, you're not wasting money, since you can use these drives to backups so stuff like this won't happen in the future.
 
i have zero experience with Linux. is this ddrescue thing bootable, and therefor it will work regardless of the OS or do i need to have Linux installed somehow for it to work?

wow this will be a pretty hefty expense buying 4 4tb drives...
is it really the only option??
and i thought the whole idea behind doing RAID 10 is to have a backup and speed in a reliable way...

Eli
 
i have zero experience with Linux.
Click to expand...

You will have to learn a little to understand how to use ddrescue. Specifically you will have to learn how to figure out what drive is the source, what drive is the destiantion and do that via the shell (command line).

and i thought the whole idea behind doing RAID 10 is to have a backup and speed in a reliable way...
Click to expand...

Common misconception. No matter what raid level you use you still need a backup for a multitude of reasons.

i have zero experience with Linux. is this ddrescue thing bootable
Click to expand...

Yes.

http://www.sysresccd.org/Sysresccd-manual-en_How_to_install_SystemRescueCd_on_an_USB-stick
 
eli_singer said:
i want to do it as safe as i can get, i just want to make sure i'm doing it right.

i have the X4 4tb drives that were in the RAID 10 array. meaning i have 8tb of data (actually it's less because they're not completely full but still, let's say it's 8tb).

i should just plug them out of the controller and in to another machine?
can i just plug them to the SATA ports on the MB?
how should i configure them in the BIOS? as RAID? as AHCI?

that will suppose to result in 4 drives recognized and accessible by Windows?

now i'm a bit confused.
going by the safe method, i'm suppose to make a 1:1 copy (with r-studio?) of the drives to some other drives? can i split this image to several 2tb drives?

and after that i should give r-studio a chance to rebuild the array on the original X4 4tb drives? am i getting this right?

thanks for all the help...

Eli
Click to expand...

the reason you're feeling a bit screwed right now is because you went below your depth when taking on a RAID array, and now you're about to drown. so either you forfeit everything and die, or you commit to becoming the expert you thought you were, and start swimming back to shore.

you can't become an expert by reading forum posts. this is a long trial you're about to begin.

I'll start you off. if you have disks that are breaking, they are as good as dead. that means trying to make an array with them is futile, because THEY ARE WHY THE ARRAY BROKE. and they could die at any time. so if you want your data back, it makes no sense to fix the array. you need working disks, therefore you need to move the contents of the broken disks to working disks. make sense?

RAID systems have at least one requirement that you didn't know of. the requirement is that you have two of them. redundant redundant arrays of independent disks. :) see where I'm going here? you need to make a second array with working disks.

can money be saved while doing this? yes. if only 2 disks are breaking, then you can get away with only buying 2 disks. image those broken disks over, and then fix the array with the 2 new disks and the 2 old disks.

in the best case scenario, you get this done. then what? you're back to where you were when you started: some bad luck away from another broken array, and another huge hassle. great. this is why they say RAID isn't a backup... unless you've got two of them, then one of them is a backup for the other.

so really, do to this right, you need to buy 2 new disks plus another 2 (as raid 0) or 4 (raid 10) to backup the WORKING array onto. either that, or just 2 disks, and use a remote backup service like crashplan. then you live with needing to download all that data (or pay to have them ship it to you) in the event of failure.

part 2 of getting you started: ddrescue is a command line tool which copies data, and r-studio is a raid array format fixer. which do you think you need right now?

ddrescue aborts copying chunks of data which are slow (dead), and then keeps track of when that happens. so a typical rescue with ddrescue results in a new disk with a bunch of blocks missing, plus a log file of the copy progress. then you re-run it with different settings to pull the skipped stuff off, and it updates the log file. you repeat this until either you get everything off, or the disk dies, or you get tired of trying and decide to live with the lost data. once you get to this point, if you want to know which files contain the unrecoverable data, a different tool must be used.

I use partedmagic to do my disk and ddrescue work. it can write to an image file, or directly to disk (the "clone" concept). I prefer to do work directly to a new disk because it removes a step and probably needs fewer disks.

once you've digested what people here are saying, you can always ask for advice on another forum for another perspective, but presenting your question from a new perspective with your new info.
 
first of all, thank you for the comments.
i do intend to swim back to shore :)
i wasn't claiming being an expert but i am learning and trying.

about the situation at hand- i understand that it is likely to assume that the drives are dead, but, they very well might not be. it may look off-topic but it might be relevant.

digging and testing a bit more revealed that the underline cause for the broken array might have to do with a bad memory controller on the motherboard and not the drives.

i got a few BSOD's with a MEMORY_MANAGEMENT error. the BSOD's went away when i took out the IBM m1015 RAID card. in that point i thought that maybe the card is faulty.
i ran memtest on my two memory sticks without the card to make sure and i got this:



i only had errors in test #6.
i tried to figure out which one is the bad stick by running memtest on each stick individually and changing slots.
i got the same results- errors in test #6.

i went out and got brand new sticks just to make sure and ran memtest on them and got exactly the same results!
this led me to believe that the problem is not bad memory sticks (obviously).
the two candidates were the CPU's L cache and the memory controller on the MB.
i ran intel tool for a while and it said my L's are fine. i don't know how conclusive that is but it does points out to the MB.
either way, CPU or MB, something is wrong there and the IBM m1015 made it worst.
adding that to the fact that all drives are pretty new (about 7 months) and how it is very unlikely that 3 drives will die one after another, i think that saying the drives are dead might be premature.

now, in light of that, i would like to believe i have better chances of retrieving the data and rebuilding the array since the drives might be ok. maybe it's wishful thinking but it's there in my head...
if that's the case, don't i have a way of making the controller import the array? or accessing the data so i can make a backup? i will still build it on new drives, but it seems logical that keeping them on the controller in an array is better than plugin them to regular SATA ports making them loose any connection to the array they were part of... isn't it?


i understand the difference between a copying solution (ddrescue) and a rebuilding solution (r-studio).
i'll take your advice and will try to use partedmagic.
if anyone can refer me to any how-to's or step-by-step tutorials about how to use this things it would much appreciated...
i have no knowledge of Linux command line syntax and such (or Linux at all as a matter of fact) so any guidance would be great :)

thanks,

Eli
 
i have no knowledge of Linux command line syntax and such (or Linux at all as a matter of fact) so any guidance would be great
Click to expand...

The blog post I linked above was reasonable.

I can give you some advice to get you started..

1. Unplug the SFF8087 cable from the raid card. Also remove all other SATA cables you have to your mobo.

2. Connect only the source drive and the destination

3. when you boot off the linux usb stick and get to the shell use the following command to find out what your drives are

Code: 
sudo fdisk -l /dev/sd?

This should return the partition table and sizes of the connected drives. Most likely if you did steps 1 and 2 you should have /dev/sda /dev/sdb and /dev/sdc. One of these 3 will be the usb stick. The other 2 will be your source and destination. It is essential that you figure out what drive is the source and what drive is the destination. ddrescue will not prompt you or care if you reversed the order meaning it will happily overwrite your raid disk with the 0s that are on the new drive it you tell it to do that...

A good way to verify the drives is by serial #. There are two easy ways to do that. hdparm or smartctl.

Assuming you detected that /dev/sda was one of the 2 hard disks from the previous step:

Code: 
sudo hdparm -I /dev/sda

Or smartctl

Code: 
sudo smartctl --all /dev/sda


Then verify the other disk ( lets assume it was /dev/sdb):

Code: 
sudo hdparm -I /dev/sdb

John
 
I forgot in step 2 connect the source and destination drives (just 2 for starters) to motherboard SATA ports.
 
One small note to add on that excellent summary:

$ sudo smartctl -i /dev/sda

will produce much less output (and scrolling) if all you need is the serial number.
 
BTW, I intentionally did not give the ddrescue command yet. The reason I would like the OP to get past the first few steps before going on..
 
thanks.
it will take me some time to do that, i need to buy 4 4tb drives and such...
you have been very helpful i i will post here with updates.

Eli
 
parted magic's main purpose is gparted, a GUI tool that shows detailed drive info by right clicking.

all the ddrescue info is in the manual, which one needs to read many times, over many days, to fully understand all of it. it's good idea to write the commands you plan to use, and an explanation of your plan, in a text file.

if your motherboard is bad, that can definitely break stuff. replace the motherboard immediately... don't keep using it unless you want it to continue breaking stuff.
 
i am replacing it of course. i am currently using a second machine to do all that. that's why it's going to take some time... at least i hope i'll have time to dive into this Linux thing...

can i use gparted to identify the drives?

Eli
 
You must log in or register to reply here.
Back
Top