ZFS preventing POST

T

tajudd

Hi guys,

I've not a whole lot of experience with ZFS. I did a search and dburke came back with as close as I can get. I did a zfsroot install as documented here. Since I had both ad0 and ad1 as identical IDE 10G drives (test drives), I followed the above RootOnZFS link on both disks. After the setup of ZFS on root, then extracting the tarballs like I always successfully do, the system cannot POST now. It gets as far as probing HDD information, then locks up. yes, locks up.

For stats reason, I did a ZFS mirror on ad0 and ad1, since I didn't have 3 physical disks to do a raidz. i386 host with 1.5GB ram.


I'm lost on how bits on a platter affect the identifying information on the HDD circuitboard.

Ideas?
 
How long have you waited? Some hardware, particularly server motherboards, takes a long time to timeout on disk detection. Or maybe it's waiting for a serial console or something similar. Check the BIOS settings.

Of course, what's on the disks should not change how the POST works... unless there's some sort of managed RAID going on, and ZFS overwrote what it was expecting.
 
it's a desktop PC, not server class in any shape. I left it for hours, and with no further progress (or timeouts). This is partly why i'm calling it locked up. the other is that a momentary push of power doesn't power down the box (soft-off)

I thought about the possibility it was due to 1 of the 2 disks that may have crapped out. I haven't tried toggling disks yet. they've been removed from the system and I haven't tried with them again. I did power cycle the PC in attempt to fix.
 
I have the same problem on a MSI K8N Neo4-FI nForce4 motherboard, can't even get in to the BIOS setup. Is your hardware the same?
 
Some broken BIOS firware might try to figure out the partitioning scheme on the drive. I've seen these kinds of problems, too. This is not a ZFS bug. This is probably broken firmware. Look for updates at your vendor's website.
 
Further to the previous response, some BIOSes seem to check for the existence of a valid MBR partition table on the boot disk, with at least one active primary partition and fail to boot if they don't exist.

The GPT standard specifies a "Protective MBR" that includes a single primary partition spanning the whole disk, to keep the real GUID partition table safe from badly written disk utilities.

With your GPT ZFS setup, you may need to mark that fake whole-disk partition as active to work around this.

I've had this problem on two systems built with Intel motherboards and flagging the fake partition as active fixed it for me. I did it with the FreeBSD fdisk utility. It complained a little but did the job.

Code: 
[root@gate] /home/jason # fdisk ada0
******* Working on device /dev/ada0 *******
parameters extracted from in-core disklabel are:
cylinders=484521 heads=16 sectors/track=63 (1008 blks/cyl)

Figures below won't work with BIOS for partitions not in cyl 1
parameters to be used for BIOS calculations are:
cylinders=484521 heads=16 sectors/track=63 (1008 blks/cyl)

Media sector size is 512
Warning: BIOS sector numbering starts with sector 1
Information from DOS bootblock is:
The data for partition 1 is:
sysid 238 (0xee),(EFI GPT)
    start 1, size 488397167 (238475 Meg), [B]flag 80 (active)[/B]
        beg: cyl 0/ head 1/ sector 1;
        end: cyl 1023/ head 255/ sector 63
The data for partition 2 is:
<UNUSED>
The data for partition 3 is:
<UNUSED>
The data for partition 4 is:
<UNUSED>

[root@gate] /home/jason # gpart show
=>       34  488397101  ada0  GPT  (233G)
         34        128     1  freebsd-boot  (64K)
        162    8388608     2  freebsd-swap  (4.0G)
    8388770  480008365     3  freebsd-zfs  (229G)

=>       34  488397101  ada1  GPT  (233G)
         34        128     1  freebsd-boot  (64K)
        162    8388608     2  freebsd-swap  (4.0G)
    8388770  480008365     3  freebsd-zfs  (229G)
 
I've seen the same on an intel board (DX975XBX2), if you somehow manage to connect hard disks without a single active partition on them the system hangs at POST.
 
My board is a cheap Biostar board. It's a basic board that boots anything (given HDD is "good"). I've downloaded the latest BIOS and will apply and test if there is any difference. Sorry for the delayed response.
 
I've never seen this issue. None of the BIOSs that I've seen require the hard drive having an active partition to get through POST. What happens if there is no hard drive? Do you need to make the partition active using a different computer or would a bootable CD or USB be enough to get through POST?
 
davidgurvich said:
I've never seen this issue. None of the BIOSs that I've seen require the hard drive having an active partition to get through POST. What happens if there is no hard drive? Do you need to make the partition active using a different computer or would a bootable CD or USB be enough to get through POST?
Click to expand...

No hard drive -- it passes and moves on just fine.
Bootable CD/USB/DVD - won't get that far.

It probes CPU and memory, skips a line like it normally does, and the HDD's are setup to "auto" in the BIOS to autodetect. This autodetection stalls/freezes the computer.


I've got the disks set aside, I just haven't tried since posting this.
 
I have had similar trouble while trying to boot GPT/ZFS USB stick too. USB sticks with proper MBR boot ok. You need to have only 'one' drive without "proper" MBR, to have it hang.
 
The only similar experience I can remember is when I installed FreeBSD on a IBM ThinkPad. Long story short; after repartitioning the hard drive and installing FreeBSD on a partition (the other still had Windows something on it) the machine wouldn't boot. Not even from a CD. Naturally, I thought the machine had broken, but when I took out the hard drive the machine was fine.

Turns out the particular BIOS version had a bug; it didn't like FreeBSD partitions. After a BIOS upgrade the problem was solved. I've installed FreeBSD on many ThinkPads after this, and have never seem this problem after that.
 
danbi said:
I have had similar trouble while trying to boot GPT/ZFS USB stick too. USB sticks with proper MBR boot ok. You need to have only 'one' drive without "proper" MBR, to have it hang.
Click to expand...

Seems to be a common thread

I'm gonna go ahead and try (keyword!) to get the MBRs zero'd out then. This is less than good because I have no hot-pluggable equipment. Gonna have to try to figure out how to get 'em zero'd.

EDIT: I have the drives zero'd out now. I used a different model of Biostar motherboard. I'll post one final one if I get these two drives in their original case and passes post. Thanks for the feedback, everyone.
 
jem said:
I've had this problem on two systems built with Intel motherboards and flagging the fake partition as active fixed it for me. I did it with the FreeBSD fdisk utility. It complained a little but did the job.
Click to expand...

My fake partitions were already marked active, but this does not resolve the issue with the BIOS. Motherboard is FOXCONN 945P7AA-8EKRS2. Not even updating to the latest 515f1p55 BIOS.
 
Issue identified

Thanks all for your feedback. Last post on this thread. I zero'd out the first 1MB of data on both disks. stuck them both back to the same previous enclosure/motherboard and the motherboard passes POST. The problem is a "BIOS Bug" and this thread can be closed. I appreciate the feedback.

Thanks!
 
You must log in or register to reply here.
Back
Top