Twin drive failure, best way to procede

[Jimbo]

Okay long time user here.

Usually I have been able to nut out issues myself by reading the forums and what not.

This time around I would like advice before proceeding.

 

I have had a twin hard drive failure.

One went red and one went orange and alll the content of both disapeared off the array.

 

I have not been able to unmount them to do any checking so I am assuming there has been some file system problem with at least one of the drives.

 

The good news is I have mounted them in my windows box and retrieved the data off them.

 

Situation is of now.

Disk 8 was 1TB and has been replaced with a new 2TB EARX dot changed from red to orange.

Disk 7 was orange, has been backed up and reinstalled.

 

I can start the array and it gives me the option to format the unformatted drives but on refresh nothing has happened.

 

I don't have unmenu currently installed.

I have never used the preclear/

I have never changed jumpers on the drives.

 

Next step before I break something?

Should I be looking at changing the drive jumpers over on the EARS and EARX's?

[dgaschk]

Since you recovered the data you can reset the array config and generate new parity. Enter "initconfig" on the command line. The 2 unformatted disks may be cleared by unRAID and the sever will be offline and uncommunicative until the disks are clear. You can use pre-clear to eliminate this down time. Forget the word "jumper". Make sure that the setting for disk format is 4K-aligned. On version 4.7 this is not the default. This setting only affects newly added drives and will not effect existing drives.

[Jimbo]

Okay here is what I have done.

 

1.installed unmenu and preclear script

2.changed block size to 4k

3.unassigned drives 7 and 8

4. initconfig

5. started array, it took along time to come up, when it did disk1 failed with errors turned red, disk 5 (I have suspected to be dying for a while) showed errors but stayed green.

6. Shut down pulled disk 1 and 5 verified I could recover data by mounting in windows box.

7. initconfig again with drives 1,5,7 and 8 unassaigned or removed.

8. rebuilt parity for remaining drives.

9. Started pre clear on new disk8

 

So it looks like I have some drive juggling to do for a week or two, I have two new 2TB EARX to replace drives 7 and 8 I have a spare unused 1TB to replace disk1 and I'll pick up one or two more 2TB depending on what the pending smart reports say about the old drives 1,7 and 5 after I get through the two new drives and move data ect.

I'm really space poor at the moment on both the server and my windows box and that will give me enough space to juggle data should this happen again.

 

One final thing.

When all is back up and running as it is. As I was experiencing performance issues before this all went down.

Would it be worth moving the data off drive 11 and then pulling it from the array and re adding it so it formats at the correct block as that would be my only EARS left?

[JonathanM]

When all is back up and running as it is. As I was experiencing performance issues before this all went down.

Would it be worth moving the data off drive 11 and then pulling it from the array and re adding it so it formats at the correct block as that would be my only EARS left?

I think it would be worth waiting to see if the performance issues were caused by the failing drive(s). I doubt a misaligned EARS would cause issues you would notice without running back to back benchmarks with synthetic data.

[Jimbo]

Fair enough.

 

Now my first 2TB has cleared without error and I have added it to the array.

 

Last night I added a 1TB EADS drive to replace disk 1 as that drive I really need back online.

Went to do a preclear on both the drive in question and another 2TB EARS.

 

The 2TB EARS preclear started as normal.

The 1TB EADS however... the screen telnet window comes up with the preclear and is stuck on 1sec no updates.

Can't cancel it with CTRL+D and if I leave it go it will lock up the server requiring a parity check.

Tried to clear with both 64 and 63 allignment.

Checked the power and sata cables in between.

 

Dud drive?

 

I am using putty and screen FYI.

[Joe L.]

Or, insufficent memory to run two preclear, or they share a disk controller and it locked up.  You did not attach a syslog, so I'm guessing you already looked there and saw no need to provide it for analysis as recommended for requests for assistance.

[Jimbo]

Didn't have one to attatch as the server locked up and I had to reset it.

 

I don't think it is a RAM issue as I re tried just the 1TB by itself twice more.

I moved the possition of the drive to a different slot and preclear started.

 

I'm starting to think I have hardware issues.

Here is the current syslog, throwing out 10B8B  DisPar errors

https://dl.dropbox.com/u/1359805/Server/syslog-2012-08-26.txt

 

System is:

P5Q Pro

2gb ram

2x Adaptec 1430SA

Intel PRO/1000

3x Norco 5-in-3 (The old ones not the SS-500)

 

Edit:

Sigh.

Preclear on 1TB just finished, stopped the array to add it in and disk6 has been disabled

Syslog

https://dl.dropbox.com/u/1359805/Server/syslog-2012-08-26new.txt