Hello list members,
hello Michael,

here ist the answer to the question: how do I upgrade from my old mdtool
raidconfiguration to the newstyle raidtools 0.90 ??

That question was asked so times, but as much as I searched wasn't answered.

This is not realy a HOWTO. It all describe here is what I've done on my
system and my configuration. I don't know wether it applies also to raid 1
or 0 configuration. I've only raid 5.
I was verry carefull not to loose any of my data. And you should do also !

As allways, I can only recommend to backup your data, just in case. (as
Murphy says: a failure will allways come, and in the worst manner)

What I had:
I was running mdtools 0.4x supplied with SuSE 6.x, Kernel 2.2.5 & 2.2.13
I had 2 raid systems with 9 Disks each, each 4 Gig. Each raid on it's own
UW-SCSI controller.
The raiddevices were filled up to 90%.

What I wanted to do:
I wanted to use the new features of the raidtools 0.90.
Upgrading without dataloss, autostart included.


At time I was doing it, I used the distribution from SuSE v 6.3 with kernel
2.2.13.
I also used follwing Raidtools and patches:  
ftp://ftp.kz.kernel.org/pub/linux/daemons/raid/alpha/raidtools-19990824-0.90
.tar.gz
ftp://ftp.kz.kernel.org/pub/linux/daemons/raid/alpha/raid0145-19990824-2.2.1
1.gz 

The kernelpatch came up with errors but it's save to ignore them (these are
allready in).

For my testing I had a very small raidsystem (9x7MB). I filled it up with
data, to see all will be all right after upgrading to new raidtools.
I could be usefull, if you some devices for testing also.

Since I have read a comment not to use the raid superblock because of
dataloss while upgrading from older raidtools or mdtools, I set it 1st to 0.

See raid0.conf. sample from archive:
persistent-superblock   0    # set this to 1 if you want autostart,
                             # BUT SETTING TO 1 WILL DESTROY PREVIOUS
                             # CONTENTS if this is a RAID0 array created
                             # by older raidtools (0.40-0.51) or mdtools!
I don't know wether it applies to raid5 also, but just to be save.

I booted with my new build kernel.
1st I had to my /etc/raidtab from my old /etc/raid5?.conf files.

2nd I had to run mkraid -- upgrade /dev/md2
No error ? Fine. Try to mount the /dev/md2, mybe ther will come a message to
fsck your filesystem. By doinig there were no errors.
Everythink worked fine, and no data was lost.

3rd I tried to the things to autorun the Raidsystem by the change of the
filesystemtype to 0xfd.
That didn't work becaus of the missing raid superblock.

So I stepped back and booted to my original configuration with the old
mdtools.
I recreated my test raid. I also set back the fstype with fdisk.
( does anybody know how to do it with only commandline options?? )

Back to my new Raidtools configuration...

1st step was done previously exept for the persistent-superblock
        I set it now to 1
2nd step: run mkraid -- upgrade /dev/md2
        all done fine again. No dataloss at all.
3rd step: do the partitiontype set to fd
        (this time I used sfdisk /dev/sde 5 -c fd) for /dev/sde5

After the boot the raid system was ready to mount :-))
I checked all data on the disk. Every little Bit was there and all in the
right place. I've had a local copy of a webserver all pages displayed at
it's best.

After that I applied these steps with my productive raidpartition.
So what do I say.
I don't need my backup. :-))


I hope that document is not as bad as I think it is, because englich is not
my language, and 2nd, I'm not used in documenting what I've done. (gives me
allway a bad commend from my boss)

Any comments an questions are welcome.
 
mfg, 
B. Schackel 
Deutsche Post AG Darmstadt 
Firewall-Admin 

Reply via email to