Re: [Testbed-admins] Booting error in Customizing the Generic Image

["'Ryan Jackson'" <rdjackso@flux.utah.edu>]

* Cheng Cui <ccui1@tigers.lsu.edu> [091101 11:01]:
> Hello Ryan,
> 
> For your previous email, yes, I am using a remote VGA console just for
> test. So the VGA mode is applicable.
> I have re-downloaded your new tarball, removed the serial terminal part
> of all the grub.cfg, and redistributed them in /tftpboot. By the way,
> the path of several kernel files including Frisbee should be changed
> under /tftpboot/, I have already modified them both within grub.cfg and
> the MySQL database, e.g. the table 'node_type_attributes'.

What did you change, and what did you change it to?

I'm not sure I understand what you were able to get working and what
you weren't.  It looks like the admin mfs works, but I am not clear on
if you were able to actually load an image onto a node or not.  Can
you clarify that for me?

> However, the booting default image failed. Please refer to my attached
> pictures. In the grub menu, it does not response to the "Emulab Boot".
> Do I miss something? Thanks a lot.
There are several different reasons that the "Emulab Boot" entry could
fail to boot.  Did this failure occur when you were trying to boot the
Frisbee MFS to load an image onto the node?  Or was this after the
image loaded when it was trying to boot it?

Try the following and send me the output:

Boot the node, and press ESC when you see "Press `ESC' to enter the
GRUB menu...".  Then press 'c' to bring up a grub command prompt.
Run the "bootinfo" command, then the "set" command and send me the
output.  After that, run the command "bootinfo_magic" and send the
output of that as a separate picture (the output of all those commands
will probably not fit on one screen).

The first command "bootinfo" queries your boss server to determine how
the node should boot.  The second command will display Grub
environment variables, and the third command uses those variables to
determine if an MFS should be booted or if we should boot from disk.

Anyway, I think we're pretty close to getting this working.  Thanks.