upthetoon Posted October 24, 2015 Posted October 24, 2015 I had a 2TB array drive fail (it was a bit of a slow death with errors and then completely failed just before I added a replacement drive). I've precleared a 3TB drive (no errors or issues) and then just followed the instructions to replace the failed drive. It all seemed fine and started the data rebuild but after only a few minutes the new drive showed up as failed too. I havent physically disconnected the old drive, I'm using different SATA and power socket/cables. I've now stopped the array. The new drive appears in the list of drives... But if I select it, it does this; Diagnostics attached, help appreciated! ridcully-diagnostics-20151024-0905.zip
garycase Posted October 24, 2015 Posted October 24, 2015 Did you change the assignment to "Not assigned" and then Start the array, so it showed the drive as "Missing" ? You need to do that; then Stop the array; and then assign the new (3TB) drive to that slot. Then when you Start the array it should do the rebuild. If that's what you did, then exactly what happened after the rebuild started? If you're referring to the status on the Dashboard, it shows the drive as bad until the rebuild completes.
upthetoon Posted October 24, 2015 Author Posted October 24, 2015 Did you change the assignment to "Not assigned" and then Start the array, so it showed the drive as "Missing" ? You need to do that; then Stop the array; and then assign the new (3TB) drive to that slot. Then when you Start the array it should do the rebuild. If that's what you did, then exactly what happened after the rebuild started? If you're referring to the status on the Dashboard, it shows the drive as bad until the rebuild completes. No, I stopped the array, reassiged the failed drive then stated the array. Damn it. It then seemed to start a rebuild but then the 5th slot showed as disabled. It seems like I should change the assignment of disk 5 to "no device" now and start the array?
garycase Posted October 24, 2015 Posted October 24, 2015 Hopefully you haven't done anything that will preclude a successful rebuild ... but this is what I'd do: (a) Stop the array and unassign disk 5 (i.e. mark it as unassigned). (b) Start the array => disk5 should now have a Red X by it and show "Missing" © Stop the array. Then assign the new drive you want to do the rebuild with (the 3TB unit) (d) Start the array ... the rebuild should start [i believe there's a box you have to check before starting the array to confirm that's what you want to happen]. At that point just wait for the rebuild. This will take a LONG time ... with your drive mix I'd estimate something in the 13-15 hour range.
upthetoon Posted October 24, 2015 Author Posted October 24, 2015 I've unassigned the drive and started the array, theres another issue, thats what's thrown me. "Unmountable disk present: Cache • WDC_WD5000AAKS-22V1A0_WD-WCAWF7406431 (sdh)" The drive has a green circle but has this message.. Partition format: error " Unmountable - No file system (32) " No idea whats happened here.
upthetoon Posted October 24, 2015 Author Posted October 24, 2015 Something weird is happening. I stopped the array and tried to assign the new drive. When I select it from the drop down is disappears from the list and the slot is shown as "unassigned" still. This new drive and the cache drive are both on the 2 SATA3 ports on the motherboard, the other drives are all on the regular SATA ports, I wonder if that's got anything to do with it.
upthetoon Posted October 24, 2015 Author Posted October 24, 2015 Thanks for your help Gary. I'm sure there's something wrong with the SATA3 ports when I use both of them for some reason. I've taken the faled drive out and re-used that port, the cache drive was then fine and the data re-build is now underway. Hopefully I'll have a fully stable server again by tomorrow! Thanks again.
upthetoon Posted October 25, 2015 Author Posted October 25, 2015 For reference if anyone finds this in similar circumstance, it was a chipset/SATA controller bug with the on board Marvell controller; http://lime-technology.com/forum/index.php?topic=40683.0
garycase Posted October 30, 2015 Posted October 30, 2015 Glad all worked out well ... and thanks for noting the issue you encountered with the controller => that may very well be useful info for someone in the future.
Recommended Posts
Archived
This topic is now archived and is closed to further replies.