Ok, it's up and running.  Some issues involved with the switch from the 
2.4 series kernel to the 2.6 series that Slack 12 uses.

I finally decided on an 8 gig partition for the OS, which leaves bags of 
room.  I did a FULL install, which is unusual for me and I'm already 
seeing packages I will uninstall, but anyway.  Oh, and FULL obviously does 
not include EMACS because I am not RMS.

The drives are sliced as follows:

   sd[ab]1        Primary  Linux swap             254.99
   sd[ab]2  Boot  Primary  Linux raid autodetect  8192.38
   sd[ab]5        Logical  Linux raid autodetect  120804.69
   sd[ab]6        Logical  Linux ReiserFS         120804.69

so sda2/sdb2 form RAID1 slice /dev/md/0 with the OS on it.  Getting it on 
was fun, but fairly straightforward, after reading this:

	http://slacksite.com/slackware/raid.html

except the system came with the new 'mdadm' tool instead of the now 
depreciated 'raidtools' used in the HOWTO. I couldn't figure out the 
'mdadm' equivilent to a 'failed-disk' entry in /etc/raidtab so I installed 
the 'raidtools' package and used what I'm familiar with.  *Then* I figured 
out how I could probably have done it with 'mdadm'...

Anyway, I installed on sdb2 and got the OS running.  Then I set the 
partition type of dsa2 to "Linux raid autodetect", and built a RAID1 
mirror on sd[ab]2 *BUT* with sdb2 configured as a failed-disk.  Then I 
brought the mirror online as md0, changed /etc/lilo.conf to use md0 (for 
boot and root) instead of sdb2 and to add the raid-extra-boot line, copied 
all of sdb2 to md0, and lilo'd with the root as md0 (had some fun there). 
Then I crossed my fingers and rebooted from sda.  Up she came, booting off 
sda and rooting the md0 mirror.  Warning:  No, you don't want to copy 
/proc and /sys but *YES* you want to create empty /proc and /sys 
mount-points.  ;-)

Obviously /proc/md0 showed the mirror degraded [U_] so I then went ahead 
and changed the partition type on sdb2 to "Linux raid autodetect", and 
used 'raidhotadd' to put sdb2 into the mirror.  /proc/mdstat immediately 
showed reconstruction going on, and I waited for that to be complete 
before altering the /etc/raidtab to show /sdb2 as no longer a 
'failed-disk'.  A reboot was in order to show that the system would come 
right up after a power-failure without any prompting, but first I wanted 
to set up LVM.  On RAID1 mirrors, of course.

I had previously set the type of sd[ab]5 to "Linux raid autodetect" so I 
used mkraid to create the md1 mirror with these two.  Then I used 
'vgcreate' to build a Volume Group called "sataraid" with md1 in it to 
give me 112.50 gig of space.  I then used 'lvcreate' to make me a 64 gig 
Logical Volume called 'music', formatted it with reiserfs.  Then I copied 
all my OGG files from sdb6 to the new /dev/sataraid/music partition, which 
I mounted under /music.

What were my OGG files doing on sdb6?  Well, unfortunately, the 
cheap-assed M/B only has support for two SATA drives.  I was forced to 
install sdb, copy the OGG files (and a bunch of other stuff) from the old 
SATA drive, take out and discard the old SATA drive, and then install the 
second new SATA drive as sda.

The current state is:

   sd[ab]1 : two 256 meg swap partitions
   sd[ab]2 : two mirrored 8 gig partitions, holding OS
   sd[ab]5 : two mirrored 112 gig partitions in Volume Group 'sataraid'
   sda6    : single 112 gig partition, empty
   sdb6    : single 112 gig partition still hosds 24 gig legacy data

The plan is to transfer the legacy data from sdb6 to LVs created out of 
'sataraid', then build another mirror as md2 using sd[ab]6 and add this 
mirror as another Physical Volume to the 'sataraid' VG.

But it's time for a beverage...


---
 þ Synchronet þ Linus is a regular at The ANJO BBS.  No, Linus Brathwaite!
--- SBBSecho 2.12-Win32