Failed drive during OCE?

[jcazes]

As I posted in the other RAID5 online capacity expansion (OCE) thread I am currently expanding my array from 4 to 6 drives on an Adaptec 51245. I say "currently" because this is an 8-10 day process...

Anyway, earlier today I hear the alarm go off on the RAID card. Apparently one of my drives failed. I think "no way," and immediately swapped it out with another unused drive in one of my hotswap bays. The Adaptec storage manager tells me the array is degraded and a few other details (none of which I can remember, given my state of panic).

After some troubleshooting, I replaced my fanout cable with a spare one I had handy, and when I rebooted the storage manager told me the array was degraded, but still "reconfiguring" (the term it used when it was doing OCE before this problem). There is also a caution notification on the drive that purportedly failed. Here is a screenshot:

If I drill down into the drive's properties, it tells me the status is Optimal (albeit with a caution sign) but that there are "failed logical device segments":

Finally, if I check out the status of the array itself, it tells me something is up with this drive (although the notifications/messages are not very clear):

At this point, I'm wondering two things:

1. How is the reconfiguration (OCE) still going on given the "failed logical device segments"? I have seen the progress on the reconfiguration continue to increment, so I don't know what's going on at this point...

2. Given these failed logical segments, should I initialize the drive and let the hot spare kick in (as the mouseover suggests)? If I do this will I jeopardize the OCE?

I will be calling Adaptec on Tuesday, but if anyone has any advice prior to that I would greatly appreciate it!

 

[Blue Fox]

Last time I used OCE, I had a drive fail. Made me pretty nervous, but that's what backups are for (I made one immediately after the drive failed...all 6TB of data ). Normally the controller will finish the capacity expansion and then rebuild the array...or at least that's how mine did it. Unfortunately it also makes OCE drag on for a lot longer than normal. Since your card takes over a week normally, I have a feeling that it will be double that now.

 

[jcazes]

Ouch, what a lengthy process! Fortunately I don't have 6TB of data to back up, though, so what little I do need to back up is already duplicated.

 

[jcazes]

Just an update for anyone searching the internet... my reconfig just finished (ahead of schedule, it seems), and sure enough the rebuild began immediately thereafter. This rebuild seems to be going MUCH faster (looks like it will take just under 2 hours), so that's a relief. Maybe this is where the faster processor on the card kicks in??

I'm still perplexed as to why the drive somehow "failed"... I just ran HDtune on it and it's OK, so hopefully it was just the cable that crapped out.

 

[jcazes]

The feelings of ineptitude are mounting!

The entire process finished (expansion and rebuilding), but I hit a roadblock... The logical array seemed to expand correctly, but I cannot expand my partition on it in Vista:

For some reason, Vista detects the additional space, but I cannot expand the volume. Here are some of the details I pulled from the command prompt:

Any ideas?

 

[Syntax Error]

Is your RAID partition set as a GPT or an MBR partition style? MBR has a limitation of 2TB for logical space, and if that's the case, expanding your partition by 900GB on top of a 1.5TB array would not be possible. I'm assuming you've checked the DISKPART utility as a method of attempting to expand your partition that way, away from Windows Disk Management, and expanding for GPT partition tables through DISKPART follows this procedure:

1. Run CHKDSK D: (or other drive letter) to make sure your file system is clean.
2. Run DISKPART from the command line.
3. Type LIST DISK to get a list of drives attached to the system.
4. Type SELECT DISK 1 (replace with the proper disk number) to select the disk you need to expand.
5. Type LIST PARTITION to get a list of partitions on that drive.
6. Type SELECT PARTITION 1 (replace with the proper partition number) to select the partition you need to expand.
7. Type LIST VOLUME to get a list of volumes mounted on the computer. Verify that there is a "*" next to the disk/partition/volume that you want to expand.
8. Type EXTEND and hold thy breath.
9. Type EXIT to get out of DISKPART.
10. Run CHKDSK D: (or other drive letter) again to make sure file system is still intact.

(Kudos to SomeJoe7777 at Tom's Hardware forums for this step-by-step list that worked for me, as well )

 

[Ockie]

You should NEVER replace a disk during expansions, you can corrupt, fail, or drag out the array expansion time greatley

 

[jcazes]

Is your RAID partition set as a GPT or an MBR partition style? MBR has a limitation of 2TB for logical space, and if that's the case, expanding your partition by 900GB on top of a 1.5TB array would not be possible.

ding ding. I lose!

Ultimately, my only option is to get 2 2TB drives, move the data, delete/convert current volume to GPT, expand, then move data back.

I did not get the memo.

 

[Ockie]

ding ding. I lose!

Ultimately, my only option is to get 2 2TB drives, move the data, delete/convert current volume to GPT, expand, then move data back.

I did not get the memo.

Or if you are lazy, you can create two partitions and create one logical volume covering those partitions.