I'm having a RAID issue and was hoping someone could clarify. At work, we're using a Sonnet Tempo SATA E4P to connect two Lacie 8TB RAID drives (4Big models) via ESATA. These two 8TB drives are mirrored via Software RAID.
How is each LaCie 4big setup?
At the moment, I presume each is a stripe set (8TB usable capacity per box), then you mirrored the pair of boxes. This is actually a 0+1 configuration, which isn't terrible, but it's not as safe as a 1+0 (aka 10).
Assuming this is the case:
- Your data is still in tact due to one of the mirrors isn't damaged (hence the Degraded state = data's there, but runs slower, and is in danger of data loss). If a disk dies in the functional LaCie box, your data = gone.
We had some issues with the power supplies failing due to the person that originally set it up. Essentially, the power supplies were all wrapped up together and stuffed into a hot corner causing them to fail.
Since then, I've ordered new power supplies and placed them in better areas. The RAID finally mounts again, but I have this error in Disk Utilities
The power loss either caused a failed write (disk is good functionally speaking, but the data's not on that particular member), or the PSU took the disk out with it.
Now don't panic - read on.
I've tried repairing the degraded RAID twice without any progress.. I guess the weird thing is the RAID Slice (E6243AE1-DCA9-4171-AD9B-68169384350B) is missing.. I'm not sure what that is or if it should even be there because we only have two drives connected.
If I understand your configuration correctly, it seems that
E6243AE1-DCA9-4171-AD9B-68169384350B is a failed member of one of the LaCie boxes (not sure which one it's in however).
Now if the disk is good (failed write due to power outage), you'll need to do the following to restore what's there:
- Get into Disk Utility and format (re-initialize) the array. The data on this particular array will be lost, but it will be usable again (stable state). As the data was mirrored on the other set, you'll still have it (why you don't need to panic).
- Once this is done, the OS X will rebuild the data on the set you just re-initialized, and once completed, everything will be fine.
If this doesn't work, the disk is bad, and do the following (not much different).
- You'll have to swap the bad disk out for a new one.
- Go into Disk Utility, and format the array (same as above; the stripe set will be empty at this point).
- Let OS X rebuild the mirror (copies the data from the good set to the newly rebuilt set; this is automatic).
Can i simply remove it from the RAID Set? I'm very concerned about losing data. Once I have this operational again, I can back this thing up and convince them to move over to a Hardware RAID system
I hope you do have a backup, and that it's current just in case. But if you do the above steps correctly, you shouldn't need to restore from the backup at all.
The LaCie products used are hardware based according to the
datasheet (pg. 2 under RAID Performance) on LaCie's site, but it's an inexpensive controller (RoC = RAID on a Chip). Though hardware however, this is certainly an argument for moving to a
proper hardware solution (do not use stripe sets), and use levels better suited to your needs (automatic rebuilds at all levels if using a nested configuration).
If they don't go for this, I'd recommend making sure you've a proper backup (needed for what I'm about to recommend), and do the following:
- Make a proper backup (no need to follow the previous information, as it will save time skipping it).
- Get to the back of each LaCie unit, and move the switch to 4 (sets it to RAID 5 + spare).
- Now go into OS X, and format both 5's you just made.
- Stripe both of the 10's you just formatted (will create a 50 with a usable capacity of 8TB, with a hot spare per box).
- Restore the data off of the backup.
Please note that this will be slower than 01, particularly due to the RoC in the LaCie's. But it's possible to speed it up a tad by setting the LaCie to 5 (= RAID 5 only) due to additional parallelism of drives. You just have to keep up with monitoring what's going and replace a disk immediately when one fails.
This will give the ability for each LaCie to automatically rebuild a degraded array (once a disk is available if you decide not to run it with a hot spare).
There's other possibilities, such as 51/61 as well (reduces your capacity, but adds additional redundancy, as if you do loose one of the 5's in a 50, the data is gone).
can you read from the 8tb cube_02 mirror?
It's in a Degraded state, but is still accessible.
