On 11/09/2011 02:57 PM, William Colls wrote: > On 11/09/2011 01:55 PM, Phil Turmel wrote: >> Hi William, >> >> On 11/09/2011 12:12 PM, William Colls wrote: >> [...] >> >>> I thought, at the time, that I needed to do the create so that the /dev/md0 device would be created properly (new machine had no raid before). >> >> That's the "--auto" option, which has sane defaults. >> >> [...] >> >>> No output from the original setup. >> >>> From what you've described so far, a likely possibility is that the original raid 1 was using metadata version 1.1 or 1.2, which put the superblock near the beginning of the disks. The default "--create" metadata in that old version of mdadm is 0.9, as you can see in your reports. >> >> If so, you've likely only lost a tiny bit of data at the end of the volumes where the 0.90 superblock has been written. (I'm also going to assume that the two disks were in-sync in old box before you moved them, so the re-sync wouldn't do any harm.) >> >> A dump of the first 8K of your drives might be helpful here. >> >> dd if=/dev/sdb count=16 2>/dev/null |hexdump -C >> >> dd if=/dev/sdc count=16 2>/dev/null |hexdump -C >> >> Regards, >> >> Phil >> > > Did the dumps to text files. diff shows no difference between the files. Should I be looking for anything particular? There are sections which seem to have data, and several blocks that are all zeros, but I don't see any particular patterns, and only a very few obvious text strings. > > Thanks for your interest and help. I was expecting them to be nearly identical, unless they're corrupted. I was particularly interested to see if an MD signature shows up at offsets 0 or 1000H, with device names at 20H or 1020H. Or there might have been a partition table or boot loader showing, or fragments thereof. I can't see them, of course. (Hint.) Phil -- To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
