| UNIX/Linux Forum Archives · Rules and Guidelines · Disclaimer |
Help
Search
Members
Calendar
|
| Welcome Guest ( Log In | Register ) | Resend Validation Email |
![]() ![]() ![]() |
| fishsponge |
|
|
Administrator ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Group: Admin Posts: 679 Member No.: 1 Joined: 13-February 03 |
I have built two RedHat 8.0 machines, both using soft-raid to mirror one 250GB disk onto the other identical 250GB disk inside the case.
During the install, I created the RAID partitions, and assigned them to actual partitions, and now I have "/dev/md0" up to "/dev/md7". However, each time I boot and each time I shut down, I get the following message: No spare disk to reconstruct array! -- running in degraded mode. What exactly does this mean? Both drives are visible from the BIOS, and the jumpers are set to Cable Select (not Master and Slave). Despite the above error message, the machine boots fine, shuts down fine, and runs fine. However, if I unplug the master hard drive, it will not boot from the secondary, even if I plug it into the primary socket on the IDE cable. I have even tried inserting a third 250GB hard drive on the secondary IDE channel to see if that would become the spare disk it needs to reconstruct the array, but that didn't work either. Does anyone know what I'm doing wrong here? Should the disks be physically set to master and slave? Do I need to do something else to the RH8.0 config to make it work? I'm really confused... Thanks in advance to anyone who can help :D |
| fishsponge |
Posted: Jan 18 2005, 03:53 PM
|
|
Administrator ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Group: Admin Posts: 679 Member No.: 1 Joined: 13-February 03 |
here is the relevant section of "/var/log/messages" which seems to contain lots of info pointing at why this isn't working, but none that helps with fixing the issue... lol
http://mongeese.co.uk/________/messages |
| fishsponge |
Posted: Jan 21 2005, 12:46 PM
|
||||
|
Administrator ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() Group: Admin Posts: 679 Member No.: 1 Joined: 13-February 03 |
ok, this issue is now resolved, and i have removed the messages file from my web server. Before i explain, please understand that all this has been done under RedHat 8.0. It may work under other flavors or versions of RH but this is what i've used. Basically, Soft-RAID only works on a partition-by-partition basis. Not on a disk-by-disk basis. Therefore, if you soft-RAID 1 everything (i.e. mirroring) then you are only mirroring the partitions.. not the disk itself. This means you have to write the MBR yourself if you wish to boot from the second hard drive. SpookyEddy from http://www.techimo.com/ gave me this link: - http://www.kieser.net/linux/raidhotadd.html and Ed Wilts from the General RedHat Mailing List gave me this link: - http://www.dur.ac.uk/a.d.stribblehill/mirrored_grub.html -------------------------------------------------- To summarise everything, the soft-raid software was reporting into /var/log/messages stuff like:
and was also reporting on shutdown and bootup stuff like:
What i basically had to do was to hot-add the partitions from the secondary disk to the raid array, and then write the MBR onto the second disk manually. I did this simply by following the instructions in the above two web pages. Also, just to confirm, if a disk fails and you insert an identical blank disk, you will have to: 1. Re-create the partitions using fdisk exactly as they are on the working disk. (Don't forget to set the partition types to be the same, and to set the /boot partition to bootable etc...). 2. 'raidhotadd' each new partition to the raid array. 3. Re-write the MBR to the second disk. Consider the issue resolved! |
||||
![]() |
![]() ![]() ![]() |