On Monday February 4, wodecki@gmx.de wrote: > Hi there, > > we had a raid0 running over here on a 2.4 kernel successfully. Today we > thought about switching from suse to debian and re-installed the machine. > Everything went fine until we wanted to start up the old raid array on the new > system. > We encountered the following error (during kernel boot, later in syslog): > > md: md driver 0.90.0 MAX_MD_DEVS=256, MD_SB_DISKS=27 > md: Autodetecting RAID arrays. > (read) hdb1's sb offset: 15012608 [events: 61ccef71] > md: invalid raid superblock magic on hdb1 > md: hdb1 has invalid sb, not importing! > md: could not import hdb1! > (read) hdd1's sb offset: 15016576 [events: ffd8773f] > md: invalid raid superblock magic on hdd1 > md: hdd1 has invalid sb, not importing! > md: could not import hdd1! > md: autorun ... > md: ... autorun DONE. Looks very much like the raid superblock is totally corrupted, or not where it should be (which comes to the same thing really). Could the drive have been reparitioned? Could one of the componenet partitions have been "mkfs"ed? Either of these would have destroyed your array. > > I tried to start, migrate, upgrade using the raidtools2 package, but I > failed deperately (yes, I've got raid0 code in the kernel, I snipped it). > As I've read through the archives and googled I found the mdctl programm, > but it doesn't help very much, neither mdctl --scan nor --examine did soething > for me: > > (linux):~/mdctl-0.5 # ./mdctl --scan > mdctl: option s not valid in mode @ > (linux):~/mdctl-0.5 # ./mdctl --examine /dev/md0 > mdctl: /dev/md0 is too small for md > (linux):~/mdctl-0.5 # mdctl --examine /dev/hdb1 would be the sort of thing that you want. --examine looks at a component to see what the raid super block looks like. --detail looks at a running array and tells you about it. However in this case, I suspect that "mdctl --examine /dev/hdb1" will just say that it cannot find a superblock. > > I hope somebody of you has got an idea what to do; I only see one way to > reconstruct the array (without loosing the data); Build a new md0 device, save > the raid superblock, copy the raw device data (yes, I saved it) back to the > drive and write the "new raid superblock" back raw on the device - is this > possible? I'm not sure. What exactly did you save? An image of the raid array (/dev/md0) or of a component (/dev/hdb1) or ... Either way, you should be able to get your data back. Let me know the details and I'll explain. Details should probably include the side of your current partitions, and the size of the image that you save. NeilBrown - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
