Hi Axel, On Tue, May 22, 2018 at 06:11:34AM +0000, Axel Spallek IT-Dienstleistungen wrote: > Hi, > [...] > I read a howto where one wrote I should recreate the raid: > Mdadm -create --level=5 --raid-devices=4 /dev/md1 /dev/sdb1 /dev/sdc1 /dev/sdd1 /dev/sde1 > That did it, but I could not mount it. > I realized, that the raid was marked degraded and tried to recover by writing to /dev /sde1. > I paniced, because all drives appeared ok (AAAA) when I scaned them so I stopped the RAID. > Then I reassembled them with the three devices but never managed to mount it. > I read that the order of the drives is important. > Is hat true? Yes. Also, the other parameters (chunk size, superblock format, etc.) have to match the "original" RAID's configuration for this to work reliably, and you have to "--assume-clean" to keep md from re-initializing the array. At any rate, that's an absolute last-ditch effort to rescue an array if _nothing_ else would. A number of (very bad) howtos jump to that conclusion prematurely, and make users lose data. > Did I destroy the RAID? Probably. DId you at least pass "--assume-clean", or did the new array initialize from scratch for a few hours? > [...] > What should I do next? Ideally, restore from backup and move on while having learnt something new. If that's not possible, wait for consensus (of more list subscribers than just me) on what to do. If you did not destroy everything with a full re-init, you could possibly salvage the array's data by `mdadm --create --assume-clean`-ing the _original_ array's exact configuration (if possible, using the same mdadm version you used when you first created that array). I hope you'll succeed at getting your data back. -- with best regards: - Johannes Truschnigg ( johannes@xxxxxxxxxxxxxxx ) www: https://johannes.truschnigg.info/ phone: +43 650 2 133337 xmpp: johannes@xxxxxxxxxxxxxxx Please do not bother me with HTML-email or attachments. Thank you.
Attachment:
signature.asc
Description: PGP signature
