• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Broken Raid5

A

alcaz

n00b
Joined
Feb 4, 2011
Messages
15
I have a Raid5 array with 7 x 2TB HDD. One drive failed and the raid started to rebuild. At 41%, it jamed and there was a timeout error count to 1 on one drive. If I change that drive, the raid won't rebuild. Instead, it show the Raid with one missing drive. Any idea?
 
I assume you replaced the failed drive and then it started to rebuild? Unfortunately, it sounds like you are screwed. I think personally that 7 drives is too many for a raid-5 array for exactly the reason you had a problem. If you power everything off and replace the drive with the timeout error count, maybe it will come back?
 
I think personally that 7 drives is too many for a raid-5 array
Click to expand...
Agreed. Use raid 6 in this case.

if the array comes back up as danswartz described (put the timeout drive back) do the following:

1. Power down
2. Make a bit for bit copy of the drive that timed out to a new drive using ddrescue (available on sysrescuecd or a recent linux livecd) with the timeout drive and a new drive connected to sata ports and the raid card or cables disconnected

3. Insert the copied drive back into the array and see if it rebuilds.
 
Last edited:
Yes I will try a bit per bit copy. I know I am going raid 6 from now on
 
WHY would you do raid 5 with 2T drives!!

raid 6 min.

7 drives isnt too much for raid 5, but 2T drives are.

if you can access the array, copy your important data off as said above and learn from this, min raid 6 with 1T + drives, especially desktop drives.
 
alcaz said:
I have a Raid5 array with 7 x 2TB HDD. One drive failed and the raid started to rebuild. At 41%, it jamed and there was a timeout error count to 1 on one drive. If I change that drive, the raid won't rebuild. Instead, it show the Raid with one missing drive. Any idea?
Click to expand...

Your array is too large for RAID 5, you need to change to RAID10 or RAID6 if your controller supports it.

It's already hard enought to make it through a rebuild. Controller failure, cable failure, power outage... and with 24+ hour rebuilds, the last thing you need is a URE during a rebuild... increasingly common with large arrays and 2+ TB drives.
 
You should mount your array in read-only mode if the remaining 6 drives at least spin up. You should only encounter a single bad file where it cant read that one block off the second bad drive.

Then copy off all the data minus that one corrupt section.
 
Unfortunately, it sounds like you are screwed.
Click to expand...
I hate to say it, but I agree with this.

I've seen this happen before, your drive probably hit a bad sector, didn't have a TLER function, and the array dropped the drive and destroyed the array. Even though it is still claiming to rebuild, all of your data is most likely lost or destroyed at this point.

For 2TB drives x7, RAID 6 is a must, especially with FakeRAID or hardware RAID.

Best of luck to you.
 
Why would it destroy the array?

I've had a URE during a rebuild. It just stops and then I was able to re-mount the array in read-only mode and copy off every file except the one that had the URE in the stripe.
 
You were very lucky that it worked out for you, it is a rarity that that happens.
 
Red Falcon said:
You were very lucky that it worked out for you, it is a rarity that that happens.
Click to expand...

Hmm, I was under the impression that that was how it usually worked.

Unless a second disk physically fails or wont show up in the system for various reasons I don't see why it cant be used in the array with whatever stripes are still in tact and readable.
 
If a failure happens, then a failure occurs during rebuild, that normally means the array is lost.
Unless you're talking about something else, I'm pretty sure that's what happens.
 
If the first failure is a physical drive death and it simply wont spin up or show up in the OS and then during a rebuild you simply encounter a URE the rebuild will stop and fail, but I don't see a reason that as long as the drive that had the URE still spins up and is readable by the OS that you can't recover every bit of data except the one that the stripe is across the unrecoverable sector.

I just using simple MD RAID and it allows this by just re-mounting the array in read-only (degraded) mode so it doesn't try to rebuild but lets you read all files. When you try to read the file that has the URE stripe then either the file reads corrupted or the OS throws a read error and refuses to copy that file depending on the software you are using.

I would only imagine a more expensive hardware RAID solution would allow this as well or else it seems really stupid for them not to when a basic free software RAID supports it.
 
With mdadm I have been able to work around UREs by

1.figuring out what drives have the URE,
2. using ddrescue to copy all the good data on these drives to new disks
3. swaping the new and old disks
4. Forcing a rebuild.

I had to do this after a power issue (sata power cable came out) knocked a 10 drive raid 6 down to 9 drive raid 5 and then there were at least 2 drives with UREs at the same time. Darn seagate 7200.10s.


Anyways the array was saved and I did not have to use my tape backup to rebuild the entire array.
 
You must log in or register to reply here.
Back
Top