EBIOS read error: 0x42 on Software RAID 0

[gxsolace]

Hi All -

Getting this error while booting Lion once in a while in RAID 0 (Apple software RAID).
It happens once in a while but I can make it go away by repairing permissions + clearing boot/kernel caches -- although I'm not sure which one is actually helping, and that may just be a coincidence.
But it'll come back, and I won't be able to boot from the RAID from that default partition.

I have a striped RAID 0 consisting of 4 SSDs, and boot from a Chimera USB boot stick. This has only happened since going to RAID -- never had any issues on single standalone drive (also booting from USB boot stick).

The odd thing is, Chimera now lists 2 to 3 identical "Macintosh HD" volumes at the boot selection screen. I only have 1 "Macintosh HD" volume on the RAID array, which is called "RAID." It also sees that too. So at the boot selection screen I see:

hd (1,3) Macintosh HD
hd (2,3) Macintosh HD
hd (3,3) Macintosh HD
hd (4,3) RAID

If I get the EBIOS 0x42 rear error, I won't be able to boot from 1 or 2, but I can boot from 3 or 4 ("RAID"). And when the EBIOS error 0x42 does happen, the listing will show up as this:

hd (1,3) <?xml versio
hd (2,3) <?xml versio
hd (3,3) Macintosh HD
hd (4,3) RAID

And so 1,3 and 2,3 will not be bootable but the others will be. Again, one RAID 0 array called "RAID" with a single volume on it called "Macintosh HD".

/gx

 

[neilhart]

Yes this is strange. Please open terminal and do "diskutil list" and post a screen capture of the terminal window.

It should look something like this capture.

neil

 

[gxsolace]

Indeed it is weird.
Here is the screen cap.

/gx

 

[gxsolace]

Thought this screenshot might help you to understand which slice belongs to which volume, since the diskutil readout may have been a bit confusing.

/gx

 

[neilhart]

Okay I see 8 60GB SSD configured into two RAID sets.

These RAID sets are named "User" and "Macintosh HD"

RAID set "User" is made up of disks # 1, 3, 5, 6 and is logical disk #7

RAID set "Macintosh HD" is made up of disks # 0, 2, 4, 8 is logical disk # 11

Disk # 9 is a 32GB drive named "Boot"

Disk # 10 is 2TB carved into partitions "Store A" and "Store B"

Disks # 12, 13, and 14 are each 2TB with single partitions; "Store D", "Store E", and "Store C"

This is a total of 9 SSDs and 4 2TB drives where your MB only has 10 internal SATA ports which means there is a secondary SATA controller or you are using some external drives.

It is not clear what boot loader you are using and if you want your RAIDs to be bootable or not. Also what version of OS X is this?

neil

 

[gxsolace]

Okay I see 8 60GB SSD configured into two RAID sets.

These RAID sets are named "User" and "Macintosh HD"

RAID set "User" is made up of disks # 1, 3, 5, 6 and is logical disk #7

RAID set "Macintosh HD" is made up of disks # 0, 2, 4, 8 is logical disk # 11

Disk # 9 is a 32GB drive named "Boot"

Disk # 10 is 2TB carved into partitions "Store A" and "Store B"

Disks # 12, 13, and 14 are each 2TB with single partitions; "Store D", "Store E", and "Store C"

This is a total of 9 SSDs and 4 2TB drives where your MB only has 10 internal SATA ports which means there is a secondary SATA controller or you are using some external drives.

It is not clear what boot loader you are using and if you want your RAIDs to be bootable or not. Also what version of OS X is this?

neil

That's all correct.
I'm using a product from a company called Angelbird (http://www.angelbird.com) that allows 4 SSDs per PCIe card to operate in RAID (Apple Software RAID). It works in a Mac Pro as well as PCs. Natively supported by OS X.

So I have a USB boot stick loaded with Chimera 1.5.4, and I'm running Lion 10.7.1. The RAID set "Macintosh HD" is bootable. Which is what I boot into when I get the boot selection screen.

Just to be clear, it does work. It's just that sometimes I would get that EBIOS 0x42 error that's only fixable by clearing boot/kernel caches over and over. Then it goes away but eventually comes back.

/gx

 

[neilhart]

I have no experience with a system with that many drives. However the EBIOS error would indicate that one of the drives is not functioning correctly (maybe intermittent). If this is very random, maybe you just live with it until the failure becomes more pronounced.

Trouble shooting a complex system is a matter of localize and isolate. As a test, remove all of the drives that you can see if the problem persists.

Also, why not make the RAIDs bootable? See InsanelyMac, vboyz posting on the topic. I am running bootable RAID 0's on three Lion systems (one with 4 slices), so it can be done.

neil

 

[gxsolace]

Indeed it is quite a complex setup I have going.

The reason I didn't install Chameleon or Chimera onto the RIAD volumes itself is because I didn't want to have to deal with putting everything on each slice's helper partition. But the primary reason is that if a Chimera/Chameleon upgrade should go wrong, it would leave the system inaccessible... whereas if my boot stick should ever have a malfunction, I can simply fix it easily.

Sometimes a bad kext can cause a KP in early boot stages and not allowing me to go in and remove the offending kext. I just felt that with a separate boot drive, problems would be easier to handle.

In regards to the possibility of a drive intermittently having a malfunction, I don't think that's the case because in the boot listing of:

hd (1,3) <?xml versio
hd (2,3) <?xml versio
hd (3,3) Macintosh HD
hd (4,3) RAID

hd (3,3) and (4,3) is always bootable without errors. Only the first 2 entries have problems, keeping in mind that all 4 represent the same volume, and in the beginning Chimera only saw one "Macintosh HD" and one "RAID." Then the other two "Macintosh HDs" popped up gradually. I ran diagnostics multiple on all 4 SSD modules that comprise "RAID" without any errors. Clearing kext/boot caches removes the EBIOS error and restores the names from "<?xml versio" to "Macintosh HD." Once that happens, hd (1,3) and (2,3) are bootable again without error. It's weird. lol.

/gx

 

[neilhart]

I understand your concerns. In my systems, I insure the ability to recover from a RAID 0 slice problem (where my RAIDs are system boot logical drives) by always having a recovery partition on another drive and TimeMachine. Also most of my systems have drives on quick disconnect slides so that I can reconfigure quickly.

However that does not address the error that you see. From my reading on the EBIOS error, the solution usually reported is repartitioning and reformatting the offending drive.
But then your use of PCIe SATA cards maybe the source of the problem(s). Tricky to trouble shoot.

When you find the source of the problem, please share your findings.

neil

 

[gxsolace]

I did consider the possibility of the PCIe SATA RAID board as the cause of the errors, but removed that possibility by moving the 4 RAID slices over to the onboard Intel ICH. They continued to work since it's just software RAID, and booted fine. Both Chimera/Chameleon still randomly show the EBIOS 0x42 error from time to time, with the only temp solution being clearing caches and rebooting.

The only thing I was able to find on EBIOS 0x42 was this: http://forum.voodooprojects.org/index.p ... ml#msg3942

Which doesn't seem related to my issue.
Still scratching my head at this one.

/gx