Koozali.org: home of the SME Server

Legacy Forums => Experienced User Forum => Topic started by: henry on February 17, 2001, 04:50:18 AM

Title: Testing RAID in v.4.1
Post by: henry on February 17, 2001, 04:50:18 AM

Hi Everybody,-

 I'm using E-Smith as a gateway for my home network with a cable connection to the internet. Everything works OK, but I lost the hard drive in the E-Smith server a week ago (one morning the hard drive was just dead) and decided to try software RAID in v. 4.1.

 It's an old P166, 40MB RAM (AT form factor, Award BIOS, supports booting from CD). I had 2 Seagate 1GB drives, attached them as masters do both IDE channels, run fdisk from a DOS boot floppy and removed all partitions (just in case). Booted from the 4.1 E-Smith CD and 20 min later it worked. When running df -k it showed 72% usage of /. It used only 258 1K blocks but the total size of the drive somehow turned out to be only 358 1K blocks. Can't figure out how this happened.

 To check RAID I disconnected the drive on the second IDE channel. The computer booted up fine - with the proper complaints about the second mirror. Switched the drives and booted up again - same story, everything is fine. On another computer removed the partition on the disconnected drive and returned it to the E-Smith server. I was expecting the mirror to be rebuilt when I boot up with both drives - no such luck. It won't see the second mirror (BIOS sees it just fine). Do I miss something?
After beeing broken once does the mirroring have to be turned on?

 If anybody has any advice/experience with those problems I'd appreciate you sharing it.

 Thanks.
 Henry.

Title: Re: Testing RAID in v.4.1
Post by: Charlie Brady on February 17, 2001, 06:05:32 PM

henry wrote:

>  To check RAID I disconnected the drive on the second IDE
> channel. The computer booted up fine - with the proper
> complaints about the second mirror. Switched the drives and
> booted up again - same story, everything is fine. On another
> computer removed the partition on the disconnected drive and
> returned it to the E-Smith server. I was expecting the mirror
> to be rebuilt when I boot up with both drives - no such luck.
> It won't see the second mirror (BIOS sees it just fine). Do I
> miss something?
> After beeing broken once does the mirroring have to be turned
> on?

Yes it does. From an early draft of a RAID recovery HOWTO being written by Peter Samuel:

    4) partition the new disk. It should be partitioned exactly
 the same as the other disk. Use the following command to
 determine the current partition details for the working disk
 /dev/hda

        fdisk -l /dev/hda

    You should see details similar to:

     Disk /dev/hda: 64 heads, 63 sectors, 1015 cylinders
     Units = cylinders of 4032 * 512 bytes

      Device Boot    Start       End    Blocks   Id  System
     /dev/hda1   *    1       131    264064+  fd  Linux raid autodetect
     /dev/hda2       132      1015   1782144    5  Extended
     /dev/hda5       132       137     12064+  fd  Linux raid autodetect
     /dev/hda6       138      1015   1770016+  fd  Linux raid autodetect

    Partition /dev/hdc using the command

        fdisk /dev/hdc

    Double check to make sure the partitions are exactly the same
    as those on the working disk, /dev/hda.

    5) Re-attach the partitions from the new disk to the raid
    devices

        /sbin/raidhotadd /dev/md0 /dev/hdc5
        /sbin/raidhotadd /dev/md1 /dev/hdc6
        /sbin/raidhotadd /dev/md2 /dev/hdc1

 6) You can see the progress of the raid resyncronisation by
    examining /proc/mdstat.

...

Some time further down the track these procedures will be integrated into the e-smith web manager, so that you can initiate and monitor RAID recovery form your browser.

Regards

Charlie

Title: Re: Testing RAID in v.4.1
Post by: henry on February 17, 2001, 09:10:53 PM

Thank you very much, Charlie.
I appreciate your quick response, even on the weekend.

Thanks.
Henry.