PDA

View Full Version : RAID 0 in Lion Failed




lbeck
Aug 3, 2011, 07:55 AM
This is strange, I have two 3TB Hitachi drives, both of them pass the disc utility verification and both appear to work fine. If I put them in a RAID 0, within 24 hours it fails. If I pull them out and re-initialize them they both work again.

Has anyone had this experience? any suggestions?



Shazaam!
Aug 3, 2011, 10:12 AM
I read somewhere that Lion should not be installed on a RAID.

Julien
Aug 3, 2011, 10:15 AM
I read somewhere that Lion should not be installed on a RAID.

I have Lion installed on a Striped RAID. You can't use FileVault though.

lbeck
Aug 3, 2011, 01:32 PM
Sorry, I should of been clearer. Lion is not installed on the RAID 0 volume, the RAID volume in question is simply for data.

It was working fine the past two days, then this morning my Mac did not recognize it and one of the drives was making a constant clunking noise. I thought for sure one drive failed. So i removed both, re-inserted and re-initialized. Now both are working fine by themselves.

I'm assuming that if I set them up in a RAID 0 again that the issue may come back. The funny thing is, I have two SSD's in a RAID 0 for other data and they work beautifully. So my first thought is that its not Lion since my other RAID volume is working fine.

But then that points to one of the disc, yet both of them work perfect on their own.

Strange. Any suggestions?

CyBeRino
Aug 3, 2011, 02:05 PM
Are you intentionally trying to lose your data?

Hard drives that make clunking noises are broken. Replace it. And use RAID 1.

spidey3
Aug 3, 2011, 02:10 PM
...one of the drives was making a constant clunking noise. I thought for sure one drive failed. So i removed both, re-inserted and re-initialized. Now both are working fine by themselves...

The drive making the clunking noise has failed. Sometimes when you let them "rest" a bit you can get them to work again for a little while, but the bottom line is it's failed, and needs to be replaced.

Spidey!!!

lbeck
Aug 3, 2011, 03:00 PM
Thanks for the response.

CybeRino ... I have multiple external backups of my data, Im not stupid. Of course I'm not trying to lose my data. Also, I don't want to use RAID 1 for this data, I need space. But thanks for giving me a suggestion without even knowing my needs. Anyway ...

The rest of this message is for Spidey ...

I have no idea which hard drive was making the clunking noise. I turned the computer off, and opened up my MP and then took them out and started over.

Both are working fine right now, I have them set up as individual disc. My hope is that the one that has the issue will show itself soon and I can send it back and get a new one.

Both also pass the hardware test. Since thats the case, I'm having a hard time believing that one of the disc is bad.

Care to elaborate on that?

CyBeRino
Aug 3, 2011, 03:24 PM
Thanks for the response.

CybeRino ... I have multiple external backups of my data, Im not stupid. Of course I'm not trying to lose my data. Also, I don't want to use RAID 1 for this data, I need space. But thanks for giving me a suggestion without even knowing my needs. Anyway ...


Well, using RAID0 and then wondering about why it might break just screams "I don't know what I'm doing". Forgive me if I had the wrong idea.


Both also pass the hardware test. Since thats the case, I'm having a hard time believing that one of the disc is bad.

Care to elaborate on that?

The hardware test does not test the hard drives in any truly useful way. Nor does it detect every reason for a RAID-0 to break.

There are many reasons a RAID-0 will break that aren't immediately apparent. The thing is, there are many reasons for a fault-tolerant RAID to rebuild one of its drives. For example, if the drives have their drive cache enabled, doing an unclean shutdown is a sure-fire way to get a desynced RAID set. It can also simply be an issue with the drive not being RAID-compatible because it allows itself to take too long to work around a media defect. You won't notice this when the drive is in stand-alone mode, but in RAID it's killing. If you're running fault-tolerant RAID, all of those reasons don't matter very much; the failed drive can be resynced or replaced (and then resynced). RAID 0 does not give you that luxury. Desynced RAID set means bye-bye data.

Long story short: RAID 0 is a bad idea in almost every situation. The only situation in which it isn't, is where it's being used as a scratch disk with data coming from elsewhere. If you need the space, set up the disks as standalone. You'll halve the chance of losing everything on the drives.

lbeck
Aug 3, 2011, 03:33 PM
Thanks for your response but I've been using RAID 0 as well as RAID 1 and RAID 10 for a long time. I understand you don't like RAID 0, but I need space and speed for the data on that volume and this is what I choose to use.

I do appreciate the response, i just don't understand your assumption that anyone who is using RAID 0 has no idea what they're doing. Or the need to tell me I should not be using a RAID 0.

Now, a failed shutdown should absolutely not cause a software RAID failure, really the only thing that should do that is a failed drive. And this is why I'm baffled because both drives are working as they should by themselves.

CyBeRino
Aug 3, 2011, 03:44 PM
Thanks for your response but I've been using RAID 0 as well as RAID 1 and RAID 10 for a long time. I understand you don't like RAID 0, but I need space and speed for the data on that volume and this is what I choose to use.

I do appreciate the response, i just don't understand your assumption that anyone who is using RAID 0 has no idea what they're doing. Or the need to tell me I should not be using a RAID 0.

You must be new on the internet :P Most people have no idea what they're doing. They just see "fast" and "no 'wasted' space" and decide it's good. It's like girls and shiny things. If you do know, then I apologize for having made the wrong (yet safe) assumption.


Now, a failed shutdown should absolutely not cause a software RAID failure, really the only thing that should do that is a failed drive. And this is why I'm baffled because both drives are working as they should by themselves.

It shouldn't, if you've disabled the drive caches. Which I'm going to guess you haven't, because to be perfectly honest I don't even know how to do it without using a real RAID controller (which will do it for you.) I'd look into that first if I were you (and the clunking didn't reappear, see below).

Also, now that you have the drives separately mounted, you can test them. Just make dd write the thing full of random data (/dev/urandom) and wait for it to either succeed or fail. Check your system logs (dmesg, /var/log/system.log) while you do it; if the drives cause I/O errors, it will be logged. If you see any such log entries, replace the drive you were testing at that moment.

lbeck
Aug 3, 2011, 03:51 PM
I understand, no worries.

What's DD short for? I'll go ahead and test the disc that way, once I know what it stands for :)

CyBeRino
Aug 3, 2011, 04:05 PM
I understand, no worries.

What's DD short for? I'll go ahead and test the disc that way, once I know what it stands for :)

I don't think anyone knows the answer to that for sure... http://en.wikipedia.org/wiki/Dd_(Unix)

one would use it as
dd if=/dev/zero of=/Volumes/disk1/file bs=1m

to write all zeroes to a file called file in /Volumes/disk1 in 1 megabyte increments.

You can also write random data using /dev/urandom, but that will take considerably longer (as I wrote the command I remembered you said 3TB...)

lbeck
Aug 3, 2011, 04:25 PM
I don't think anyone knows the answer to that for sure... http://en.wikipedia.org/wiki/Dd_(Unix)

one would use it as
dd if=/dev/zero of=/Volumes/disk1/file bs=1m

to write all zeroes to a file called file in /Volumes/disk1 in 1 megabyte increments.

You can also write random data using /dev/urandom, but that will take considerably longer (as I wrote the command I remembered you said 3TB...)

Gotcha, doing it now

lbeck
Aug 4, 2011, 03:01 PM
Well, there was no need to test the disc. Last night the culprit disc showed its ugly self and completely crashed. Glad I know which one it is, already have a new one being sent out.

Thanks