On Sat, Dec 22, 2012 at 03:24:53PM +0100, Volker Armin Hemmann wrote: > Am Samstag, 22. Dezember 2012, 13:53:42 schrieb Alan Mackenzie: > > Hi, all.
> > Just built kernel 3.6.11 and when I tried to install it with lilo, I got > > this difficult error message: > > Fatal: Trying to map files from unnamed device 0x0000 (NFS/RAID > > mirror down ?) > > . So I eventually had a look at dmesg for my raid setup, and found this > > - note lines 15 - 19: > > [ 2.148410] md: Waiting for all devices to be available before > > autodetect > > [ 2.149891] md: If you don't use raid, use raid=noautodetect > > [ 2.151546] md: Autodetecting RAID arrays. > > [ 2.180356] md: Scanned 4 and added 4 devices. > > [ 2.181819] md: autorun ... > > [ 2.183244] md: considering sdb6 ... > > [ 2.184666] md: adding sdb6 ... > > [ 2.186079] md: sdb3 has different UUID to sdb6 > > [ 2.187492] md: adding sda6 ... > > [ 2.188884] md: sda3 has different UUID to sdb6 > > [ 2.190484] md: created md6 > > [ 2.191883] md: bind<sda6> > > [ 2.193224] md: bind<sdb6> > > [ 2.194538] md: running: <sdb6><sda6> > > 15 [ 2.195855] md: kicking non-fresh sda6 from array! > > 16 [ 2.197154] md: unbind<sda6> > > 17 [ 2.205840] md: export_rdev(sda6) > > [ 2.207176] bio: create slab <bio-1> at 1 > > 19 [ 2.208520] md/raid1:md6: active with 1 out of 2 mirrors > > [ 2.209835] md6: detected capacity change from 0 to 34359672832 > > [ 2.211187] md: considering sdb3 ... > > [ 2.212444] md: adding sdb3 ... > > [ 2.213691] md: adding sda3 ... > > [ 2.215117] md: created md3 > > [ 2.216349] md: bind<sda3> > > [ 2.217569] md: bind<sdb3> > > [ 2.218765] md: running: <sdb3><sda3> > > [ 2.220025] md/raid1:md3: active with 2 out of 2 mirrors > > [ 2.221231] md3: detected capacity change from 0 to 429507543040 > > [ 2.222508] md: ... autorun DONE. > > [ 2.230821] md6: unknown partition table > > . Further perusal of a log file showed this error first happened on > > 2012-11-29. It would appear /dev/md6 has been firing on one cylinder > > ever since, and I've been unaware of this. :-( > > What does it mean for sda6 to be "non-fresh"? > > /dev/md6 is my root partition (including /usr :-(), so I can't unmount > > it for investigation. > > Could somebody please suggest how I might go about repairing this > > problem. > boot from systemrescuecd > mdadm -S /dev/md6 > mdadm -A /dev/md6 This didn't quite work, since mdadm -A merely restarted the array without the non-fresh partition. Still it got me searching, and what eventually worked was mdadm /dev/md6 -a /dev/sda6. (Where -a stands for "add".) The mdadm man page is very vague for this use case. > get some coffee. Make some popcorn. The resync will take some while. Indeed it did. The coffee settled me down somewhat. Thanks again! -- Alan Mackenzie (Nuremberg, Germany).
