With a RAID1 (Mirror), if one disk fails the other should run the array in degraded mode. As there was no critical data on the array I was in the process of testing various scenarios, one of which was a critical absolute failure of a disk, simulated by booting the machine without the drive.
That's when I ran into OpenE not booting with the "Unknown Operand" error, and have so far failed to both resolve it, and figure out what's causing it. The testing continues.
what is happening is that open-e writes a serial number on the disk to keep track of them
when you broke the raid 1 by removing the disk, the software was expecting a new disk drive to be added back to the raid
by adding the just removed drive you are confusing the software.
to keep you testing going, by reformat the drive so that there is nothing on it.
(I know there is no data but the serial number is there)
you can use the console utility in CRT-ALT-X "delete contants of unit"
this will wipe the drive clean.
also I think the software RAID does not support hot swap of the drives.
(at least it did not when I tested this )
Thankyou for the feedback Symm, some updates are warranted I think!
- I am installing onto a 4Gb USB pen , "FAT" filesystem (Fat16) with the default chunk size. A friend of mine who uses OpenE from a USB pen uses a 2Gb drive, and when you do an openE install to a HDD it creates a 2Gb partition. I wondered if this was significant.
- Initially, I powered down the machine, then removed the disk. It was not a "hot" removal. OpenE then failed to boot up, when you would _expect_ it to boot up, but to indicate that the affected array was now in degraded mode. However, it crashed on boot with an Unknown Operand error. Had that been a real scenario, with a disk losing power or plain old dying outright, I would have been locked out of my storage server and in a right mess! Lucky for me that I discovered this in testing, not in deployment.
- Installing to a HDD has been so far successful, with the system booting up with 1 disk removed from a 4-disk array. This leads me toward there being a problem with either the USB pen in use, the machine's USB bus, or the fact that I'm using a 4Gb partition on the USB pen instead of 2.
Having installed OpenE onto a spare 20Gb PATA disk I am now seeing the expected behaviour and no more Unknown Operand errors.
This has, at least, indicated that the problem lies somewhere between my CPU and USB pendrive. Whether it is the pendrive, the partition size, the USB bus on my motherboard... I am unable to say.
Does anyone have any recommendation on the sector size I should format to when I format the pen drive? I have tried the darned thing in USB1.1 and USB2.0 mode, FAT32 and FAT16, still bouncing into errors.
I have ordered a 2Gb drive just to test that theory also (Only £5, so worth a shot), if that one gives me the Unknown Operand trauma, then I'll have resign myself to adding another HDD to the system to run OpenE, or perhaps look to an IDE DOM.
Well, I have re-installed with a Sandisk 2Gb "Cruzer Blade" (A lovely little thing, only £5 and tiny! From 7dayshop in the UK) and it appears to be working perfectly!!
Score -1 for any memory stick from HP, though. I can't confirm if it was due to the 2Gb / 4Gb remark from earlier, but it just would not play nicely.
The 2Gb Cruzer Blade is doing great so far, will update this thread if I run into more problems.
Member
Posting Permissions
Reply With Quote