mdadm and lvm removed hard drive
Fedora core 10 - 18.104.22.168-117.fc10.x86_64
mdadm - v22.214.171.124
I have removed a hard drive from my system (broken) which was /dev/sdb. I have an array set up with mdadm using /dev/sd[cdef]1 but now that I have removed the drive the device names have all shifted. (sdf=sde, sdc=sdb, etc...)
Now when I mount my array it starts using 3/4 drives.
I tried to add /dev/sdb1 to the array with
mdadm --manage /dev/md0 --add /dev/sdb1
That works but it wants to rebuild the array (because that is basically what I was doing with that comand).
Is there a way that I can have the drive names changed and have no effect on the array, or just not have the drive names changed or some way in which I may avoid this problem in the future?
my /proc/mdstat file:
> Personalities : [raid6] [raid5] [raid4]
> md0 : active raid5 sdb1 sdc1 sde1 sdd1
> 1465151808 blocks level 5, 64k chunk, algorithm 2 [4/3] [_UUU]
> [=>...................] recovery = 6.0% (29483812/488383936)
> finish=190.9min speed=40048K/sec
> unused devices: <none>
my /dev/mdadm/mdadm.conf file: () new  removed
> DEVICES /dev/md0 (/dev/sdb1) /dev/sdc1 /dev/sdd1 /dev/sde1 [/dev/sdf1]
> ARRAY /dev/md0 level=raid5 num-devices=4 metadata=0.90
Yes, edit /etc/fstab. I hope you are using device names rather than uuid's, because that can be a pain in the arse.
With fstab, you can pretty much set the boot order and all, but mdadm (that is, software RAID) can be a serious problem if you like to sleep. Also, if you are striping, you can't recover.
I would install a real RAID card and not use software RAID, because it can really cost you, especially if you are striping. The uuid is the real problem. Even if you try to use "uuidgen" it isn't going to match.
I haven't tried it yet, but I have been thinking about skipping uuid and simply going the old way with ex. /dev/hda1 /, /dev/hda2 /var, etc.
I think that security has gone too far if I can't control my own farm, and I am in control of a lot of gig's.
|All times are GMT -5. The time now is 01:50 PM.|