Site menu:

Sponsored by

Bitcube Ltd.  Expert Linux Consultancy

Currently...

  • @popey superb, now you can cook and iron your shirts at the same time. Quite what your colleagues will think of the hickory aroma I dunno in reply to popey 1 week ago
  • @geehowquaint surely normal people are never "in" fashion, only weirdos :-) in reply to geehowquaint 1 week ago
  • not bad hold music on this teleconference 1 week ago
  • ripped jeans with backing cloth to avoid bare skin does seems a little silly. I doubt I'll ever understand fashion. 1 week ago
  • Dear PCWorld, I knew your computer skills were bad, I didn't know until recently that your grammar was too. It's not "a 3gigs memory". 1 week ago
  • More updates...

Categories

Meta

Site search

 

April 2007
M T W T F S S
« Mar   Jun »
 1
2345678
9101112131415
16171819202122
23242526272829
30  

Archives

Links:

FakeRAID - avoid

Well fakeraid really is a pile of unpleasantness. I need fakeraid since it’s the only way Vista will be protected against hard disk failure short of buying a hardware raid card (which have their own problems).

Linux support for fakeraid is a bit crap. The installers don’t handle it at all - debian and ubuntu both require you to manually set it up (and provide your own copy of dmraid to do so). Debian at least then gets horrifically confused between what is an array and what is a disk. So basically it’s manual debootstrap process.

Then you have to fight grub to install that.

But at least then it works right? Well no, because “degraded array” isn’t understood by dmraid. So if the array is degraded I have to boot into Vista to fix it :-( I can’t even _boot_ linux otherwise.

One of my hard disks has been dropping out of the array, so I’ve RMAd it. However today when I rebuilt on it, Vista BSOD on me. What state did it leave the array in? Well it marked the good drive as “bad”. (The bad drive wasn’t marked since it had dropped out).

So the PC _refuses_ to boot it from the good drive since it’s marked bad. But the _really_ bad drive it will try and boot from, but Vista just BSODs on that…

Many hours later, I find http://iswraid.sourceforge.net/mpbutil-0.2.tar.gz“>mpbutil from the ancient http://iswraid.sourceforge.net/“>iswraid site. This tool tries to mangle the bits on the disk to set them back to “ok” state. Not for the fainthearted. After reading the code so that I can see what it’s doing (in a _horrifically_ ugly manner), I give it a go. Hmm, why is dd taking so _long_ to seek to the position near the end of the disk …. ah, it’s too large - oh, it has a bug in it, which meant that if the disk isn’t recorded in the disk anymore, then it tries to write to -1 (a big number when it’s an unsigned integer). Fixed that and hey presto - disk back on line (I’d removed the dodgy disk permanently before doing this).

I’ll be documenting this, if not tidying up that code in case I need it again. I’ll also be asking Intel for the specs so that I can at least have a look at getting linux to support degraded fakeraid arrays.

Write a comment