[prev in list] [next in list] [prev in thread] [next in thread] 

List:       npaci-rocks-discussion
Subject:    Re: [Rocks-Discuss] Compute-0-0 hangs on pxe exit
From:       "Dominic Daninger" <domd () nor-tech ! com>
Date:       2012-05-29 22:43:35
Message-ID: 000c01cd3dec$7b97e220$72c7a660$ () nor-tech ! com
[Download RAW message or body]

We have observed similar problems with Intel Romley based Sandy Bridge CPU
nodes, the items below fixed the problem on at least one series of Intel
made HPC server boards. There is a problem with compute nodes booting from
their local drive after the original PXE install of the Rocks compute node
image. The following message appears on the compute node and then the node
hangs:

Trying to load: pxelinux.cfg/0A01FFFC                     ok

Booting from local disk..

 

This appears to be the fix: for the problem explained on this link:

http://www.syslinux.org/wiki/index.php/Hardware_Compatibility#LOCALBOOT_on_I
BM_x3850_X5

 

http://en.wikipedia.org/wiki/SYSLINUX 

 

You can do this:

 

# cp /usr/share/syslinux/chain.c32 /tftpboot/pxelinux/

# rocks add bootaction action=os args="hd0" kernel="com32 chain.c32"

# rocks list bootaction

 

To verify that it worked:

 

ACTION            KERNEL                RAMDISK                 ARGS
install:          vmlinuz-6.0.2-x86_64  initrd.img-6.0.2-x86_64 ks

ramdisk_size=150000 lang= devfs=nomount pxe kssendmac selinux=0 noipv6
ksdevice=bootif nonm biosdevname=0 install headless: vmlinuz-6.0.2-x86_64
initrd.img-6.0.2-x86_64 ks

ramdisk_size=150000 lang= devfs=nomount pxe kssendmac selinux=0 noipv6
headless vnc ksdevice=bootif nonm biosdevname=0

memtest:          kernel memtest        -----------------------

----------------------------------------------------------------------------
---------------------------------------------

os:               com32 chain.c32       ----------------------- hd0

 

 

pxeflash:         kernel memdisk bigraw pxeflash.img            keeppxe

 

 

rescue:           vmlinuz-6.0.2-x86_64  initrd.img-6.0.2-x86_64 ks

ramdisk_size=150000 lang= devfs=nomount pxe kssendmac selinux=0 noipv6
rescue ksdevice=bootif nonm biosdevname=0

 

 

Now all nodes that use the "os" bootaction will have this. If you find that
your heterogenous compute environment doesn't allow for all nodes to use the
chain.c32 command (not sure why they would, does anyone know?) then you can
add a different bootaction that can be assigned on a per host basis.

 

Dom

 

-----Original Message-----
From: npaci-rocks-discussion-bounces@sdsc.edu
[mailto:npaci-rocks-discussion-bounces@sdsc.edu] On Behalf Of prasad
karulkar
Sent: Tuesday, May 29, 2012 4:54 AM
To: npaci-rocks-discussion@sdsc.edu
Subject: Re: [Rocks-Discuss] Compute-0-0 hangs on pxe exit

 

 

hi

 

do you mean its goes upto PXE bootloader and then give error as exiting PXE
bootloader ?

 

if so then its not booting from PXE your frontend might not installed
properly.

 

 

prasad

 

 

 

> Date: Sat, 26 May 2012 12:43:50 -0400

> From:  <mailto:abs0lutek0ld@gmail.com> abs0lutek0ld@gmail.com

> To:  <mailto:npaci-rocks-discussion@sdsc.edu>
npaci-rocks-discussion@sdsc.edu

> Subject: [Rocks-Discuss] Compute-0-0 hangs on pxe exit

> 

> Hello all,

> 

> I've got a pair of IBM eserver 345's for testing that I've just 

> dropped 6.0

> (i386) on. The install on the frontend went fine, insert-ethers found 

> the compute node and it seemed to install fine but when it rebooted it 

> got as far as exiting the PXE bootloader and compute-0-0 hangs hard. 

> Ctrl-C and Ctrl-Alt-Del give no response the only way to do anything 

> is to powercycle the node. I've tried cutting out the unmanaged hub 

> and just using a crossover cable. I've forced a reboot and reinstall 

> to no avail. Both machines are identical and the frontend reboots with 

> no issues. Has anyone had a similar problem or any ideas?

> 

> Thanks in advance

> Andrew Gahan

> Nuclear Engineering and Radiological Science Undergrad University of 

> Michigan

> -------------- next part -------------- An HTML attachment was 

> scrubbed...

> URL: 

> <https://lists.sdsc.edu/pipermail/npaci-rocks-discussion/attachments/20>
https://lists.sdsc.edu/pipermail/npaci-rocks-discussion/attachments/20

> 120526/541dc9c9/attachment.html

                                                                            

-------------- next part --------------

An HTML attachment was scrubbed...

URL:
<https://lists.sdsc.edu/pipermail/npaci-rocks-discussion/attachments/2012052
9/5383b4bf/attachment.html>
https://lists.sdsc.edu/pipermail/npaci-rocks-discussion/attachments/20120529
/5383b4bf/attachment.html 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://lists.sdsc.edu/pipermail/npaci-rocks-discussion/attachments/20120529/5c612246/attachment.html \



[prev in list] [next in list] [prev in thread] [next in thread]