Fresh 7.0 Install: Fatal Trap 12 panic when put under load

[Kris Kennaway]

John Sullivan wrote:
>  
>> John, a question, how is swap set up on your system?  I was 
>> swapping to a file (a memory disk device /dev/md0).  I was 
>> doing this because for some reason lost in ancient history, 
>> this machine was not set up with a real swap partition.  
>> Hence, no crash dump.
> 
> Swap is a partition on the 1st disk.
> 
>> Last night I repartitioned a second disk, set up a real swap 
>> partition and now I'm currently waiting for this to happen 
>> again so I can get a crash dump.
> 
> I will try creating a swap partition on my second drive to see if that improves things ... I am able to cause a panic "on demand"
> but a crash dump is rarely written (presumably because the system believes the device is not accessible?).  I must have crashed it
> 10-20 times now  with various corruptions of the panic screen - once it had blue text with "trap 12 trap 12" all over the screen, I
> liked that one ;-).
> 
> I did manage to complete a "make index" while the background FSCK was running, once it had finished, performing the same task caused
> a panic locking the machine up again with no crash dump.

OK, the first thing to do is disable bg fsck, then force a full fsck of 
all filesystems.  bg fsck does a poor job of fixing arbitrary filesystem 
corruption (it's not designed to do so, in fact), and you can get into a 
situation where corrupted filesystems cause further panics.

Removing KDB_UNATTENDED from your kernel will allow you to interact with 
the debugger and obtain backtraces etc, which is useful when dumps are 
not being saved.

Kris