Koozali.org: home of the SME Server

RAID 1 Recovery (6.0b3)

Rob (Microtrader)

RAID 1 Recovery (6.0b3)
« on: November 08, 2003, 12:23:00 AM »
Sorry if this has been posted before, but search didn't reveal exactly what I was looking for.

Anyhow, I'm playing with the software mirroring in 6.0b3.  I did a fresh install on a pair of 6GB IDE drives using the RAID1 software mirror option.  I waited a bit and confirmed that the drive was properly in sync.

I then proceeded to simulate a failure of the first HDD /dev/hda (unplugged the power).  While not everything instantly switched over to the remaining drive (i.e. I couldn't access the server manager) , with a reboot I was back up and running on the single drive.

Next, I shutdown the sytem, switched the "working" drive to be the primary master and put in a blank 6GB drive into the system.  I booted up, ran fdisk, created the partitions to be exactly the same (type raid-auto detect and set the boot flag on the first partition).  I did the raidhotadd utility for md0, md1, md2 and everything synced up again on the "replacement" drive.

Here's the problem.  When I try to simulate a second faiilure and boot of the drive that I manually re-synced it doesn't boot at all.  It won't even show the "mytel networks" screen for the boot loader (the screen stays black with a blinking cursor.

Did I miss something?
Thanks in advance.

Rob (Microtrader)

Re: RAID 1 Recovery (6.0b3)
« Reply #1 on: November 08, 2003, 12:33:54 AM »
Sorry, I meant mitel not mytel (stupid spelling mistake)

Dan Brown

Re: RAID 1 Recovery (6.0b3)
« Reply #2 on: November 08, 2003, 12:44:07 AM »
On the black screen with a blinking cursor, do you see the letters LI immediately before the cursor?  If so, take a look at /etc/lilo.conf.  There will be a line that starts with "boot=", which probably reads "boot=/dev/hda".  It should, instead, read "boot=/dev/md0" (or md1, I forget which--a search here on RAID and lilo.conf should find more information).

Rob (Microtrader)

Re: RAID 1 Recovery (6.0b3)
« Reply #3 on: November 08, 2003, 01:09:36 AM »
No "LI"  Just a flashing "_"

If I plug in both drives again and look at /etc/lilo.conf it's already

"boot=/dev/md0"

It almost seems as if the boot loader isn't getting mirrored to the second drive (or at least not properly).  But there is something in the boot sector since it doesn't just say "no operating system found" when I boot the machine.

Rob (Microtrader)

Re: RAID 1 Recovery (6.0b3)
« Reply #4 on: November 08, 2003, 01:13:34 AM »
Update:  Fixed it.  Your post just inspired me to try running lilo and that made it happy.

Thx

Duncan Drury

Re: RAID 1 Recovery (6.0b3)
« Reply #5 on: November 29, 2003, 05:23:35 PM »
What exactly did you do to fix it?  I am having the same problem (well, getting LI and then it hangs on boot), but have found lilo.conf has boot=/dev/mda0

Any help appreciated.