[[ Boy, this list is taking a long time to get expanded ! ]]
>> Did you ever move a primary to a logical ?
> No. And I just tried to. It doesn't do it. And if I think about it, it
> would be quite "hard" to accomplish this.
I can't think why ...
Basically you have to decide the start address of the source, and the start
address of the destination, and copy from src to dst, starting from the lowest
numbered block if dst < src, and starting at the highest numbered block if
dst > src.
logical and physical partitions don't matter ....
But yes -- PM doesn't work like that :-(
>> 1.5) run resize2fs on md1, while still running Linux.
>> This "should" work for RAID[0134], but not linear ...
> I didn't know that it is possibble to shrink a partition with Linux tools :-)
Sure is -- the only way to do it !
> Nevertheless, PM is really easy to use
... if you like rebooting, and using a GUI ...
I prefer to be able to do it all from home ...
> and has really done good jobs so far.
Sure, but it doesn't talk RAID, so what you are planning to do "should"
work (as they are binary copies), but I'd prefer not to rely on it !
> According to http://lwn.net/980514/a/resizefs.html the resize2fs seemed
> to be implemeted initially in PM 3.0, so 4.0 probably has the same thing
> with a different GUI.
resize2fs is a Linux command line tool to *just* reduce the size of an ext2
partitions.
However, its great advantage is that it runs on Linux, so works with whatever
block device you care to give it, e.g. /dev/md* (RAID[0145]).
>>> 2. Boot the server from a DOS Disk with Partiton Magic ("PM").
>> Personally, I'd use fdisk rather than Partiton Magic, as you can do it all
>> from Linux without rebooting until the deed has been done (to re-read the
>> partition table, which needs a reboot -- real bummer that !!!!).
> ...thought I had to because otherwise I couldn't shrink a partition without
> loosing all data.
Nah -- as per "1.5)", shrink it using resize2fs first ...
>> If md1 is RAID1, it stands some chance of working anyway ...
> Yes, it is. It is a patched 2.0.36 Kernel (raid0145-19990128-2.0.36) with
> raidtools-19990128-0.90, running RAID1.
NB "stand some chance" -- I'd not risk it ...
>> 1) backup all data
>> 2) dismount and raidstop md1
>> 3) run resize2fs on md1
>> ensure there is at least 8K free at the end of each partition
>> for the PSB (it's 4K on a 4K boundry)
> Can I run resize2fs after md1 has been stopped? Shouldn't I just umount it?
Sorry -- yes -- I was wrong -- I incorrectly merged steps ...
2) unmount /dev/md1
3) run resize2fs on md1
3.5) raidstop md1
