Raid5 w/ mdadm request for help.

chris lostpkts at gmail.com
Fri Sep 5 15:06:56 UTC 2008


/* cross posted to forums as well */

I have an issue that I'm hitting a brick wall on and hope that someone
here can help me out.

I have 2 servers.   Hoth v2  and Hoth v3   v3 being the new better
faster server. :)

Hoth v2 = Gutsy
Hoth v3 = Hardy

Brought up v3 and added in the new hard drvies for the array  8x500
with 1 more for a spare (9 total).

rsynced the data from v2 to v3.

After 4 days of rsync... I did the following.

o removed the partitions on the drvies in v2.
o shut the box down and gutted it.
o put the 8x 400 w/ 1 spare (9 total) from v2 itno v3.

Turned on v3.  v3 FREAKED.  wouldn't bring up /dev/md0 which was the
original 500s it was built with.  Figured out that it was b/c drive
assignments changed.  Fixed that

v3 sees the follwoing for the 400s.

root at hoth:~# cat /proc/mdstat
Personalities : [linear] [multipath] [raid0] [raid1] [raid6] [raid5]
[raid4] [raid10]
md1 : inactive sdi1[0](S) sdp1[8](S) sdo1[6](S) sdn1[5](S) sdm1[4](S)
sdl1[3](S) sdk1[2](S) sdj1[1](S)
      3125669888 blocks


they all spares.

so I did a


mdadm --examine --scan /dev/sd[ijklmnop]1

then did a

mdadm --assemble --uuid=<string> /dev/md1

Said not enough devices.  So tried this.

root at hoth:~# mdadm --assemble --run /dev/md1
mdadm: failed to RUN_ARRAY /dev/md1: Input/output error
mdadm: Not enough devices to start the array.


dmesg is showing the follwoing now..


[  412.949638] md: md1 stopped.
[  412.949649] md: unbind<sdi1>
[  412.949653] md: export_rdev(sdi1)
[  412.949693] md: unbind<sdp1>
[  412.949696] md: export_rdev(sdp1)
[  412.949713] md: unbind<sdo1>
[  412.949715] md: export_rdev(sdo1)
[  412.949743] md: unbind<sdn1>
[  412.949746] md: export_rdev(sdn1)
[  412.949762] md: unbind<sdm1>
[  412.949764] md: export_rdev(sdm1)
[  412.949780] md: unbind<sdl1>
[  412.949782] md: export_rdev(sdl1)
[  412.949798] md: unbind<sdk1>
[  412.949800] md: export_rdev(sdk1)
[  412.949815] md: unbind<sdj1>
[  412.949818] md: export_rdev(sdj1)
[  413.053861] md: bind<sdl1>
[  413.053978] md: bind<sdn1>
[  413.054097] md: bind<sdo1>
[  413.054245] md: bind<sds1>
[  413.054393] md: bind<sdj1>
[  413.054417] md: kicking non-fresh sds1 from array!
[  413.054422] md: unbind<sds1>
[  413.054425] md: export_rdev(sds1)
[  413.054428] md: kicking non-fresh sdo1 from array!
[  413.054431] md: unbind<sdo1>
[  413.054434] md: export_rdev(sdo1)
[  413.054436] md: kicking non-fresh sdn1 from array!
[  413.054439] md: unbind<sdn1>
[  413.054442] md: export_rdev(sdn1)
[  413.054444] md: kicking non-fresh sdl1 from array!
[  413.054447] md: unbind<sdl1>
[  413.054449] md: export_rdev(sdl1)
[  413.079463] raid5: device sdj1 operational as raid disk 1
[  413.079467] raid5: not enough operational devices for md1 (7/8 failed)
[  413.079514] RAID5 conf printout:
[  413.079516]  --- rd:8 wd:1
[  413.079517]  disk 1, o:1, dev:sdj1
[  413.079518] raid5: failed to run raid set md1
[  413.079554] md: pers->run() failed ...


so now


root at hoth:~# cat /proc/mdstat
Personalities : [linear] [multipath] [raid0] [raid1] [raid6] [raid5]
[raid4] [raid10]
md1 : inactive sdj1[1]
      390708736 blocks



What I'm wanting to do is to wipe md1 off the earth and rebuild it
from scratch with /dev/sd[ijklmnop]1

I tried to --zero-superblock --force /dev/md1  but that didn't work either.

root at hoth:~# mdadm --zero-superblock --force /dev/md1
mdadm: Unrecognised md component device - /dev/md1


Can someone please help?

Thanks




More information about the ubuntu-users mailing list