I have been trying to get this software Raid set up with my new SuSE 8.2 machine but it doesn't appear to work. YaST2 won't let you RAID 2 partiions together and label it /boot. So I leave it out and set the raid up and then try to test it but the only way I can find to test it is to unplug the master drive and see if it boots. Which it doesn't it just goes to a flashing prompt which leads me to belive the software raid isn't working.

Here is how I have it set up. What should the boot loader properties be?

hda1 swap
hda2 RAID
hda3 RAID
hda4 RAID

hdb1 swap
hdb2 RAID
hdb3 RAID
hdb4 RAID

md0 hda2 hdb2
md1 hda3 hdb3
md2 hda4 hdb4

Help would be great. I have been trying to get this set up and tested for awhile and it I just can't firgure it out.

Thanks,

Not knowing anything about SuSE - did you read the LVM HowTo?

Håkan

No I haven't but I will if you think it will help.

I have setup RAID1 + SuSE both non-bootable on an existing install and bootable on a new install. My notes are attached to a post on justlinux.com (search raid + suse). I will post them here as well if someone can tell how to turn on "post attachments".

browns1977, you are correct that YaST will not allow raid for a /boot partition. Yast will allow raid for / partition (so boot from /). Yast will give a warning that LILO may fail in this confuguration. I ignored the warning and had no problems. Thus far the bootable raid appears to work fine but I have not yet tested the raid by simulating a disk failure. I will do a simulated disk failure test in the next week or so and will repost if I have any problems. If you set up raid + suse and test disk failure please post your result.

I downloaded your document you suggested. It looks like I am doing it per your instructions except that I haven't been raiding my swap. I have just made a swap on each drive. Should I raid them?

I am going to test it the way you suggest I have done it that way but I am going to follow your directions and see if the results are different. But I haven't has a successful test my way. Thanks for the information. And if you have anything else to offer about testing and the swap space I would appreciate it.

browns1977, you can swap on raid if desired.

Consider this quote from the Software-RAID-HOWTO section 2.5: "Another reason to use RAID for swap is high availability. If you set up a system to boot on eg. a RAID-1 device, the system should be able to survive a disk crash. But if the system has been swappping on the now faulty device, you will for sure be going down. Swapping on a RAID-1 device would solve this problem."

I am not an expert on RAID or the best way to allocate a file system over disk partitions. Refer to the various HOWTO's and the System Admin Guide for further info (www.tldp.org). P.S. I do plan to swap on RAID myself.

brucehohl,

Thanks for your help and the paper that you wrote. I am happy to see I was on the right track. I have the swap in a Raid now. Now I am back to testing. Did the testing you suggest work for you? The way you suggest is the only way that I have found to do this. Everyone suggests it but with varied result. Also in your set up you have hda and hde, I am assuming you have hda set as the master correct? My system doesn't seem to like it when the power is removed from the master.

Thanks again,

browns1977, I have not yet had time to simulate disk failure. You noted that the configuration from my worksheet passed the disk failure test for the disk b (my disk e) but did not pass the disk failure test for disk a.

May I suggest the following. While disk a is unplugged (simulated failure), boot from SuSE CD1 (rescue system) or from a boot disk in order to gain access to disk b (my disk e). Mount /hdb to /mnt. Open /etc/lilo.conf with vi and switch the hda's with the hdb's (my hde's). Save and then try to reboot. Let me know if this works or let me know if you find another or better solution. (We need a RAID expert to weigh in.)

I assume that this problem only relates to booting. This is, if disk a fails while the system is up, the system would stay up on disk b and all would be OK -- until it was time to reboot.

brucehohl,

I am testing and will try what you suggest. Still trying to get hdb to mount. I agree that this seems like it is only a problem with booting. I have been trying to figure out if this is a problem with the system bios.

Testing this Raid is turning out to be more of a pain that the actual Raid setup. My main intent is just to see how things will work if there is a problem with this system and the Raid.

Maybe we should try posting a new topic with the subject "Calling all Software Raid Experts"

I will keep testing. Hopefully we can figure this out.

Thanks,

News from my new tests.

I installed setting up the raids and all. Then testing your idea that the raid would hold and booting would be the issue if the primary drive failed, I changed lilo.conf to use hdb instead of hda. Turned the system off, unplugged hda, and then rebooted. Same result - goes to a black screen with a flashing cursor. I then looked in the bios and found that everything was fine meaning that it was looking to the slave to boot from and all. Therefore I have ruled out bios being a reason for failure.

Maybe I am overlooking something but doesn't this mean nothing is being copied to hdb? Maybe I am just getting frustrated and I am overlooking something silly.

In order to pass the failed disk tests for both disk 1 and disk 2 the solution on my test system is as follows. I edited the Yast generated lilo.conf by removing:
boot=/dev/hda
and adding:
boot=/dev/md0
raid-extra-boot="/dev/hda,/dev/hde"

Run # lilo. On my test system lilo generated some warnings which according to an internet search were related to my older BIOS. I ignored these warnings without problem. After this change I was able to complete the disk failure test for both disk 1 (hda) and disk2 (hde).