thinking about building a RAID.. HW questions - Page 3

{snicker}
Your assumption is that I'm not right and superior...

You may feel dumb and wrong, but that's not what I really think of you. I am trying to help you Earthling, FCS.

If you intend simply to mount the entire array at a single location, I'd say not to worry about LVM. LVM is a great solution if you expect to be resizing "partitions" in the future. If you're just going to use all the disk space in the array as, say, /home, then don't bother using LVM. Just format /dev/md0 with the filesystem of your choice and mount it.

I've found I can use an array while it's still in the rebuilding phase, but it's pretty slow. If you run "top" from a command prompt, you'll probably see that rebuilding process at or near the top of the list. If you "cat /proc/mdstat" you'll see a progress meter and a speed report on the rebuilding process. You might try to speed it up with "nice". In top, as root, you can hit "r" and pick a process to "renice." Larger negative numbers (up to -20) allocate more resources to a process. Try a value like -10, then view /proc/mdstat and see if it matters. My guess is that building an array is more bound by disk speeds than by CPU resources, so nice may not matter.

You can't test performance until the array is completely built. Wait until tomorrow and test it again.

RAID Error during POST. Help!

I have the Gigabyte GA-E7AUM-DS2H MB using onboard RAID 5 and it looks like I had a bad power connector on one of the three 1TB Seagate hard drives in my HTPC. The drive then caused a RAID error. I replaced the power connector but I still had an error in POST while booting up but at least now it shows all three with errors instead of two. I also have a 250GB main drive that is not in the RAID system which is fine.

I reloaded Vista with the Gigabyte RAID 64 driver again. Still the same thing. While booting up it flashes three red errors for the three 1GB drives although I think it's reading three separate (1,2,3) 1.8TB errors. When I enter Vista I do not see the three drives in My Computer or the storage manager but if I load the Nividia software I can see the 250GB drive and the three 1GB drives although it lists an error next to each 1GB drive. I am able to run the SMART test on all three drives and it reads HEALTHY for each of the three drives.

The bios is set for RAID and during boot if I press F10 to look at the RAID setup it shows all three 1TB drives with an error next to all three. Everything seems to be set up correctly and the drives seem healthy.

What can I do to fix the error and get this running again? It seems like all the drives are working and healthy but I can't fix the RAID error.

I don't know if you realize this or not, but you just posted that in the linux forum, so the first thing we're going to do is tell you to install linux.

Sorry! Missed that. I was just trying to find a knowledgeable RAID forum and saw this one looked good. Never mind!

seriously, Hogweed75...you could use one of the live CD's such as puppyLinux that 'might' let you a look at the drives without writing anything to them. If that works, you could back them up.

Interesting! http://www.puppylinux.org/ Thanks.

I was thinking that he might be able to rebuild the raid that way, but chances are slim since it's a bios raid and a pretty new chipset. It's worth a shot though, Hogweed75, if you're still around.

I'm booting from a WD 250GB for the main drive which is not included in the RAID array. This has Vista 64 and all programs on it. The three 1TB are the storage array. Hoping to not have to rebulid and lose whats on the drives but I'm wondering if that's what it's coming to. It just seems like it's a setup issue or something. Like all the drives are healthy and I just need to reset something. I hope!

P.S. I can't per say, mark the drive but I could remove it from the set up and then reinsert it and rebuild it I guess. If I remove it would I need to format it and then reinsert it in the array to rebuild. Sorry but I'm completely new at this and I've only got about a month of actual hands on under my belt in the HTPC/storage realm.

This may be helpful if you're able to get in there --> http://linux-raid.osdl.org/index.php/Linux_Raid

Correct, RAID 5 so this should work. So that I don't take any chances, what is the proper procedure?

Enter the RAID setup during POST and delete the drive from the array. Reformat the drive? Put the drive back into the array and rebuild the system?

I've run into an issue with my 5x500GB raid 5.. I've been ignoring it but need to get working on it now. I have a raid0 and another raid5 that is working fine on the same pc..

When I stress the array at all the drives quit responding and the file transfer freezes. Usually happens when transferring large HD files from this raid 5 array to another drive/array or when transferring through samba. The system doesn't freeze up but the transfer itself does and I can't do anything else with that mounted filesystem (or unmount it). I have to reboot to get the drives back up. It doesn't seem to happen on files less than 5GB or so.

ubuntu 8.04 64 bit
Drive array is running XFS no lvm. I've checked all the drives using smart tests to make sure they are not going bad and they pass all the tests without issue. The drives are in a sata backplane that seems to be working ok.. any possibility the backplane is responsible? System logs are clean as well. Any one have any ideas how to figure this out?

I don't think you have to format it, but I'll leave that up to the experts here.

Well unless anyone has anymore advice on this I think I'll try this first. Delete the drive from the array save and exit, reboot and add the drive to the array and then select rebuild array and see what happens.

Not bad for a bunch of Linux guys!

Really...Thanks for the advice on this even though I'm running Windows. I'll try it out tonight and see what happens and let you know.

It was a similar situation (drive problems) that introduced me to Linux, a much earlier version of Puppylinux was suggested to me as a way to get into a drive that windows refused to boot.

And then curiosity got me hooked because it ran so well.

It is a great thing to know and be conversant in, even if you prefer windows.

I'm planning for next years project, a large RAID storage drive most likely using Unraid which I think uses Linux doesn't it? One way or another it looks like I'll being learning Linux at some point with the different projects I keep looking at. I have some very good software engineer friends here at work that may become even closer once I take on the task! Of course they only touch Windows when they have to.

Thanks again for everyone's help today. I'm anxious to get home and try this tonight.

if you jump in next year, the btrfs file system should be out, stable and ready for use. Much better option than unraid. Basically it's the same as ZFS from solaris but less buggy

Morning Update!
Got it all working last night and it was actually very easy. Thanks for all the advice. It really helped! Some of it was just my fear of screwing it up since I'm new to working with any kind of RAID.

During POST I hit f10 and entered the RAID setup. I first removed the faulted drive from the array. I then tried to just reinsert it into the array but it doesn't let you. So then I removed the other two also. I then just rebuilt the array by selecting the three drives and saving. If it matters, I did make sure they were all selected in the same order that they originally were in. I thought I try it this way before trying to add and rebuild the drive.

Upon reboot POST had no more errors and showed one 1.8TB drive. There was rejoicing! Windows posted that there was an error and I needed to insert the Vista disc. I set up to repair Windows and it finished, rebooted and loaded Vista without a problem. I checked My Computer and there she was. 1.8TB drive with all my movie files intact. Tested several movies and everything is running perfectly!

Thanks again to all of you for all of your help even though it's Windows. I really appreciate it!!! It's always funny to me with Linux guys. Just like my engineers around here at my work. I'll ask them a Vista question and after they grumble a bit about using Windows and especially Vista and how they haven't used it for so long and don't remember this or that. Then they start looking at it, tell me to try this or that, punch around on one of there PC's and then give me a better answer that anyone else that I try asking. It seems that the people using Linux have already fought the bugs on Windows and switched therefore they are usually the best ones to ask.

I just posted a script to monitor RAID arrays in Linux over in this thread. Just thought some of you might find it helpful.

Edit:I found the bug in the script referenced below. I used a double equals sign as a comparison operator in one place; replacing it with a single equals sign fixed the problem. (I write a lot in PHP which uses "==" in if() statements routinely.) It should work correctly now.

Sorry!



I'm sorry to hear that, Scott. I wrote it to monitor CentOS boxes, but I thought it was pretty generic. If you have a chance, send me the errors in a PM, and I'll take a look.

If you just want to start the monitoring, add the line:

/sbin/mdadm --monitor --mail=you@example.com /dev/mdX &

to /etc/rc.local, replacing X with the number (0, 1, ...) assigned to each array. You'll need one line for each array. Replace you@example.com with the address to which notices should be sent. This approach will launch the monitor(s) upon reboot, but if anyone fails for some reason, it won't be restarted. The script I wrote handles that case by checking periodically when it's run from cron.