On Tue, May 22, 2018 at 06:11:34AM +0000, Axel Spallek IT-Dienstleistungen wrote: > I read a howto where one wrote I should recreate the raid: https://unix.stackexchange.com/a/131927/30851 > Did I destroy the RAID? Maybe. It depends... If the RAID-5 was in sync before, and used all of the same offsets, and you re-created with the same number of drives... then the only thing that is really damaged is whatever the metadata overwrote (if you used a different metadata version before). For all other data, XOR is XOR is XOR, even if you use the wrong chunk size and all, the result is still the same. That is assuming it was in sync before, of course. Example with 4 loop devices: # truncate -s 100M a b c d # losetup --find --show a /dev/loop0 # losetup --find --show b /dev/loop1 # losetup --find --show c /dev/loop2 # losetup --find --show d /dev/loop3 Create RAID 5 with chunk 64: # mdadm --create /dev/md42 --chunk=64 --level=5 --raid-devices=4 /dev/loop0 /dev/loop1 /dev/loop2 /dev/loop3 mdadm: Defaulting to version 1.2 metadata mdadm: array /dev/md42 started. Put some data on it: # shred -v -n 1 /dev/md42 shred: /dev/md42: pass 1/1 (random)... # md5sum /dev/md42 85febad8a4c2bd97b91e4da882d44b60 /dev/md42 Now destroy the RAID, re-create with wrong chunksize, wrong drive order: # mdadm --stop /dev/md42 mdadm: stopped /dev/md42 # mdadm --create /dev/md42 --chunk=512 --level=5 --raid-devices=4 /dev/loop3 /dev/loop0 /dev/loop2 /dev/loop1 mdadm: Defaulting to version 1.2 metadata mdadm: array /dev/md42 started. With this, the data seems to be gone. # md5sum /dev/md42 01d6d4ce2af6c4d7891aefc96efb9152 /dev/md42 # expected 85febad8a4c2bd97b91e4da882d44b60 Re-create RAID with original drive order and chunk size: # mdadm --stop /dev/md42 mdadm: stopped /dev/md42 # mdadm --create /dev/md42 --chunk=64 --level=5 --raid-devices=4 /dev/loop0 /dev/loop1 /dev/loop2 /dev/loop3 mdadm: Defaulting to version 1.2 metadata mdadm: array /dev/md42 started. # md5sum /dev/md42 85febad8a4c2bd97b91e4da882d44b60 /dev/md42 Data is back. So a wrong re-create does not necessarily harm your data. But it depends. If the RAID wasn't in sync before, or you re-created in a way that will cause the on disk data to actually change from the original layout, then it might not be possible to do a recovery anymore. Whatever you do, use overlays for recovery experiments. (This is mentioned in the StackExchange answer I linked above.) It would help if you had mdadm --examine output of your original RAID layout, otherwise you can just trial & error, or try to deduce correct RAID structure from raw data. Regards Andreas Klauer -- 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
