GEOM Raid3 drive array question

[cbowlby]

Hi Guys,

I'm running a FreeBSD 8.0 system with an "OS" drive (mirrored), and 3 other HD's in a RAID3 array via GEOM. One of my drives has gone bad, and I'm waiting on a delivery of disks (only a few hours out) so I can swap the disk out and rebuild the array.

As this is the first time I've had to do this, I am simply looking for confirmation and/or advice on ensuring I don't mess the array up when I swap the disks.

First however, some details. The array consists of 3 x 500 Gbyte disks:

ad8, ad14 and ad16.

dmesg kicks out this set of messages:

ad8: FAILURE - READ_DMA48 status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=965661086
GEOM_RAID3: Request failed (error=5). ad8[READ(offset=494418476032, length=1024)]ad8:
GEOM_RAID3FAILURE - READ_DMA48 status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED>:  LBA=931254672Device dataarray: provider ad8 disconnected.

ad8: FAILURE - READ_DMA status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=0
ad8: FAILURE - READ_DMA status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=0
ad8: FAILURE - READ_DMA status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=1
ad8: FAILURE - READ_DMA48 status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=976773167
ad8: FAILURE - READ_DMA status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=128
ad8: FAILURE - READ_DMA status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=16
ad8: FAILURE - READ_DMA status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=0
ad8: FAILURE - READ_DMA status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=512
ad8: FAILURE - READ_DMA status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=128
ad8: FAILURE - READ_DMA status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=16
ad8: FAILURE - READ_DMA status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=0
ad8: FAILURE - READ_DMA status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=512
ad8: FAILURE - READ_DMA status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=64
ad8: FAILURE - READ_DMA status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=0
ad8: FAILURE - READ_DMA status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=2
ad8: FAILURE - READ_DMA status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=16
ad8: FAILURE - READ_DMA status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=128
ad8: FAILURE - READ_DMA status=71<READY,DMA_READY,DSC,ERROR> error=4<ABORTED> LBA=0

Telling me that ad8 is the failed disk, based on those details.

a quick look at status of the array confirms the array is degraded:

test# graid3 status
           Name    Status  Components
raid3/dataarray  DEGRADED  ad14
                           ad16

and listing some more details:

test# graid3 list
Geom name: dataarray
State: DEGRADED
Components: 3
Flags: NONE
GenID: 1
SyncID: 1
ID: 1006246370
Zone64kFailed: 16281
Zone64kRequested: 183522
Zone16kFailed: 269778
Zone16kRequested: 1892123
Zone4kFailed: 25205
Zone4kRequested: 576030
Providers:
1. Name: raid3/dataarray
   Mediasize: 1000215723008 (932G)
   Sectorsize: 1024
   Mode: r1w1e1
Consumers:
1. Name: ad14
   Mediasize: 500107862016 (466G)
   Sectorsize: 512
   Mode: r1w1e1
   State: ACTIVE
   Flags: DIRTY
   GenID: 1
   SyncID: 1
   Number: 1
   Type: DATA
2. Name: ad16
   Mediasize: 500107862016 (466G)
   Sectorsize: 512
   Mode: r1w1e1
   State: ACTIVE
   Flags: DIRTY
   GenID: 1
   SyncID: 1
   Number: 2
   Type: PARITY

Confirms that ad8 is the failed disk.

Once the drives get here I will be swapping out ata4, which is where ad8 maps hardware wise. As I've never done an array disk swap and rebuild I wanted to confirm the following steps, and or get corrections/advice in advance.

1. unmount the array, and hotswap the disk.
   
   
2. execute "graid3 remove -n 0 dataarray"
   
   based on ad8 mapping to number 0 in the array (any way I can confirm that before hand? it's possibly it's number 3, but "graid3 list" is not providing those details).
   
   
3. execute "graid3 insert -n 0 dataarray ad8" (assuming the new disk is remapped to ad8)

Do I then need to execute rebuild before remounting, or does the insert automatically do that for me?

At this point is it safe to remount the array, or should I leave it unmounted while the array rebuilds?

Do I need to "pre-label" the new drive before inserting it into the array?

Should I load/unload the array at any point, thus far I've not fond a crisp clean "how-to" or doc online that goes through that in detail (could be my search terms, but "geom raid3 freebsd rebuild array" gave me clues, but nothing concise).

I apologize in advance if this is not in the right forums, but if anyone can clear up my "list of operations", I would be greatly appreciative.

 

[cbowlby]

Alright, I feel a little red in the face atm, seeing as the array already took the drive out of the array. All I had to do was hotswap the disk, then execute:

graid3 insert -n 0 dataarray ad8

The array is currently syncing to the new disk.