freebsd stable - Apr 2004 - FreeBSD 4.9 / Supermicro 7043P-8R / Crashes After 2-5 Minutes Of Uptime

Hello

I have a brand spanking new Supermicro 7043P-8R server with dual Intel
3.2gHZ Xeon processors and 4GB of Kingston memory.

I installed FreeBSD 4.9 on the box and it gives me the following message on
the screen about 2-5 minutes after it finished booting:

boot() called on CPU#0

I get this message no matter whether I leave the console alone or if I log
in and try to work on the machine.

I read a thread about a problem very similar to this from November of 2003,
but the thread seems to have just died without any resolution.

Is this a known issue with Supermicro Motherboards?  Does anyone have any
suggestions as to a potential patch or other fix?

I'm going to start doing the hardware swapping thing in a bit and see if
that fixes anything, but I'd really like to hear back from anyone who has
any experience with this issue.

Thanks a million!

Tim

Tim Gustafson
MEI Technology Consulting, Inc
tjg@meitech.com
(516) 379-0001 Office
(516) 480-1870 Mobile/Emergencies
http://www.meitech.com/

On Fri, 2 Apr 2004, Gustafson, Tim wrote:
> Hello
>
> I have a brand spanking new Supermicro 7043P-8R server with dual Intel
> 3.2gHZ Xeon processors and 4GB of Kingston memory.
>
> I installed FreeBSD 4.9 on the box and it gives me the following message on
> the screen about 2-5 minutes after it finished booting:
>
> boot() called on CPU#0
This means the machine is trying to reboot for some reason.  Typically,
its due to a panic.  You should get a lot more output with a message and a
traceback, or if you have ddb compiled in, a db> prompt.

If you aren't getting anything, try setting up serial console and log to
another machine.
> Is this a known issue with Supermicro Motherboards?  Does anyone have any
> suggestions as to a potential patch or other fix?
>
> I'm going to start doing the hardware swapping thing in a bit and see
if
> that fixes anything, but I'd really like to hear back from anyone who
has
> any experience with this issue.
Random panics are generally caused by bad memory, CPU cache, anod other
hardware issues.


-- 
Doug White                    |  FreeBSD: The Power to Serve
dwhite@gumbysoft.com          |  www.FreeBSD.org

From: Doug White [mailto:dwhite@gumbysoft.com]
> On Fri, 2 Apr 2004, Gustafson, Tim wrote:
> 
> > Hello
> >
> > I have a brand spanking new Supermicro 7043P-8R server with 
> dual Intel
> > 3.2gHZ Xeon processors and 4GB of Kingston memory.
> >
> > I installed FreeBSD 4.9 on the box and it gives me the 
> following message on
> > the screen about 2-5 minutes after it finished booting:
> >
> > boot() called on CPU#0
> 
> This means the machine is trying to reboot for some reason.  
> Typically,
> its due to a panic.  You should get a lot more output with a 
> message and a
> traceback, or if you have ddb compiled in, a db> prompt.
> 
> If you aren't getting anything, try setting up serial console 
> and log to
> another machine.
> 
> > Is this a known issue with Supermicro Motherboards?  Does 
> anyone have any
> > suggestions as to a potential patch or other fix?
> >
> > I'm going to start doing the hardware swapping thing in a 
> bit and see if
> > that fixes anything, but I'd really like to hear back from 
> anyone who has
> > any experience with this issue.
> 
> Random panics are generally caused by bad memory, CPU cache, 
> anod other
> hardware issues.
FYI, i'm using the same system. I would guesstimate that you
have a problem running out of ram (sounds silly, but yes) due
to the large amount of ram you have.

Here's what i have in mine, currently running 5.2
kern.maxswzone=16777216
kern.vm.kmem.size=314572800
kern.ipc.nmbclusters=16384
kern.ipc.nmbufs=65536
kern.ipc.shm_use_phys=1
net.inet.tcp.tcbhashsize=16384
kern.ipc.maxsockets=32768
kern.maxfiles=34000

but i have run 4.7 and releng_4 on it.

Definitely i would suggest running memtest86 on it for a bit [at
least 24 hours], but with the ECC, memory errors would have to
be gross to be noticeable.

> Definitely i would suggest running memtest86 on it for a bit [at
> least 24 hours], but with the ECC, memory errors would have to
> be gross to be noticeable.
Don,

Thanks for your response, and thanks to Doug White for his response too.

I got no other information on a serial console or on the regular VGA console
other than syncing disks and the usual shutdown messages.

I removed all memory and tried new memory in both 2GB and 4GB amounts.  I
have removed both processors in turn.  Neither of these helped at all.

I downloaded a copy of 5.2.1 and installed it onto the machine and it seems
to be operating smoothly, for the moment.

Once I was able to keep the machine running under 5.2.1 for more than 5
minutes, I downloaded and installed memtest and ran it.  It is behaving a
little oddly.  It will run if I specify 4M of memory, or 128M of memory, but
gets an immediate page fault if I specify 512M, 1G, 2G or 4G of memory.
Coincidently, it gets NO errors when run under 4M or 128M.

I suspect this might be because I have an "old" 5.2.1 kernel and
userland.
So, just to be sure, I am now cvsuping my /usr/src and I'm going to do a
make (build|install)world and I'll rebuild the kernel for good measure as
well, just to be sure.

Once all that is done, I will re-run memtest and see what happens.

Thanks for your suggestion!

Tim

Tim Gustafson
MEI Technology Consulting, Inc
tjg@meitech.com
(516) 379-0001 Office
(516) 480-1870 Mobile/Emergencies
http://www.meitech.com/

From: Gustafson, Tim [mailto:tjg@meitech.com]
> Sent: April 7, 2004 11:46 AM
> To: 'Don Bowman'
> Cc: 'freebsd-stable@freebsd.org'
> Subject: RE: FreeBSD 4.9 / Supermicro 7043P-8R / Crashes After 2-5
> Minutes Of Uptime
> 
> 
> > Definitely i would suggest running memtest86 on it for a bit [at
> > least 24 hours], but with the ECC, memory errors would have to
> > be gross to be noticeable.
> 
> Don,
> 
> Thanks for your response, and thanks to Doug White for his 
> response too.
> 
> I got no other information on a serial console or on the 
> regular VGA console
> other than syncing disks and the usual shutdown messages.
> 
> I removed all memory and tried new memory in both 2GB and 4GB 
> amounts.  I
> have removed both processors in turn.  Neither of these helped at all.
> 
> I downloaded a copy of 5.2.1 and installed it onto the 
> machine and it seems
> to be operating smoothly, for the moment.
> 
> Once I was able to keep the machine running under 5.2.1 for 
> more than 5
> minutes, I downloaded and installed memtest and ran it.  It 
> is behaving a
> little oddly.  It will run if I specify 4M of memory, or 128M 
> of memory, but
> gets an immediate page fault if I specify 512M, 1G, 2G or 4G 
> of memory.
> Coincidently, it gets NO errors when run under 4M or 128M.
i would use the 'memtest86', which runs as its own operating
system. http://www.memtest86.com/

Let me know if you want a version which understands the ECC
on this board and will load from FreeBSD loader. I sent
my patches to the author but didn't see them incorporated.

--don

Don,

Thanks for all your help here.

I downloaded the memtest86 program you suggested and it ran for 24 hours
without a single error.  I'm positive, at this point, that it's not a
hardware issue.  There must have just been something wrong with that
particular installation of 4.9-STABLE.

I'll keep you posted if anything else happens with this.

Thanks again!

Tim

Tim Gustafson
MEI Technology Consulting, Inc
tjg@meitech.com
(516) 379-0001 Office
(516) 480-1870 Mobile/Emergencies
http://www.meitech.com/