Greg Oster wrote:
Really? We've had something similar happening to us a while ago. The
system was running on a 3 disk RAID 5 array. A supermicro backplane went
up in flames[1] and the server was shut down. An engineer booted the
server without the backplane, but with one disk missing from the RAID 5
array, so upon next boot the missing disk was connected and a rebuild
was ordered.
However, the rebuild was taking too long and the server was rebooted to
do the rebuild in the background. The server came up fine and we were
searching for the necessary tools to start the background rebuild. We
thought we were in the clear when suddenly windows started acting all
weird, eventually crashed (I didn't see if it was a bluescreen, we were
working via terminal server at that time), and the server was rebooted
again. After this incident, the controller configuration was gone and
all three disks appeared as "Ready". Gone was the RAID.
We eventually were able to recover some data by re-creating the array
(luckily someone knew the blocksize originally used) and not
initializing when reconstructing the RAID. The operating system was
foobar though and couldn't be booted anymore.
This was with an intel-rebranded LSI card with an intel BIOS.
Anyone got any similar experiences with hardware RAID cards? Hardware
RAID has always been misery for me.
[1]
http://users.opengate.be/~glenn/album/index.php?folder=/Misc/Burnt%20Sup...
| Andrew Morton | -mm merge plans for 2.6.23 |
| Chuck Ebbert | Why do so many machines need "noapic"? |
| Bart Van Assche | Integration of SCST in the mainstream Linux kernel |
| Greg Kroah-Hartman | [PATCH 023/196] MCP_UCB1200: Convert from class_device to device |
git: | |
| David Miller | Re: [BUG] New Kernel Bugs |
| Jarek Poplawski | Re: [PATCH] pkt_sched: Destroy gen estimators under rtnl_lock(). |
| Gerrit Renker | [PATCH 31/37] dccp: Remove manual influence on NDP Count feature |
| Gregory Haskins | [RFC PATCH 00/17] virtual-bus |
