Wednesday 19 October 2016

Day 733, Rubbish RAID, rubbish disk


Marvellous.  Definition.  Something which has had any reason to marvel about it removed.

For some reason when I passed on one of my old PC's to someone else I configured it with a mirrored pair of disks using the on-board Marvell RAID controller.  I was obviously thinking at the time that this would save effort in the event that a disk failed.

Ha ha, what a thinking ahead smartypants I am.

Ha ha ha, what an idiot.

I wasn't counting on how piss-poor the RAID controller actually was - I'm used to enterprise grade RAID controllers that, among other things, let you know a disk has failed and can be used to rebuild arrays in a simple and time-efficient manner.

The Marvell RAID controller doesn't let you know a disk has failed other than at some point, an inexplicable time later, the PC refuses to boot as the BIOS decides to disable the on-board RAID BIOS.  This happens some time after the disk has failed, not on the first reboot.  It also gives no indication that there is a fault at the startup screen.

Re-enabling the RAID BIOS from the motherboard BIOS allows the PC to boot - sometimes.  It also enables access to the RAID BIOS utility (CTRL-M at startup), which would be good if it was of any actual use.  The RAID BIOS showed a single disk with no indication that there was a disk missing other than the state of the array being DEGRADED.  I worked out which disk had failed and replaced it with one of the multiple spares I'd left in the PC chassis.

The replacement disk was recognised in the RAID BIOS.  Unfortunately there was no option to rebuild the array, which was odd.   I could configure the single added disk as RAID 0, which was of no use whatsoever.

With the replacement disk added to the array controlled the PC then refused to boot.  I disconnected the disk, booted the PC, then plugged in the SATA cable after startup... It's a SAS disk so theoretically it should be possible to 'hot add' - your mileage may vary and I cannot be held responsible for you breaking anything, anywhere, at anytime, or indeed killing yourself due to a poorly earthed PSU.

I downloaded the Marvell RAID Utility from some place on the internet - easier than it sounds, the Marvell website was not in any way helpful.  The utility showed the working disk and the added disk.  However the added disk also appeared to be broken in some way as even though it was identified correctly there was no way to assign it to the array or even as a spare.  I had a further two spare disks in the chassis - this machine had had a large RAID array configured previously.  The next disk was recognised by the RAID Utility in Windows and did start to rebuild, although it failed shortly after 13% and then the disk was marked as 'unplugged'.

The final disk was connected.  This rebuilt completely.

Then shortly after the rebuild this disk was marked as 'shut down'.  There was no way to bring it back.  Rebooting into the BIOS RAID utility the new disk was marked as SPARE, which seemed hopeful, however there was still no way to change the status of the disk in what is supposedly the main configuration source for this type of RAID array.

A restart of the PC showed the disk still as 'shut down' and checking the properties of the disk showed that 'the disk is not well' and there would be risk if it was used.  So all three spare disks had given up the ghost, which is odd as they'd been sitting still for quite some time, they were all working previously and should have had plenty of life in them.

Bollocks.

"What did you do during your day off?"

I'm afraid I spent a substantial amount of it dicking about with a really crap array controller and some really crap disks, other than that it's been fine...

It's not as if there were pleasant, non-IT related things to do.



Argghgghhhhhhhhh.

Arrrrrrggggggggggggghhhhhhhhhh.

Arrrrrgrrgrgrgrgrgrgrgrgrgrrrrrggggggggghhhhhhhhh.


For official/internal use only:
7474
0-9

No comments:

Post a Comment