[Quantian-general] 7.9.1 and 2.4.27 kernel
Dirk Eddelbuettel
edd at debian.org
Tue Jan 3 12:29:01 UTC 2006
Hi Chris
On 2 January 2006 at 19:22, Christopher Heiny wrote:
| For some reason, the mount point /mfs does not seem to be created when
| booting with
| linux24 lang=en
| which prevents OpenMosix from starting up correctly. This happens at
| least some of the time, and maybe always. Is this a known problem, and
| is there a cure?
Yes, I think I've seen it too, and I am not sure why it happens. There is
explicit code in /etc/init.d/knoppix-autoconfig:
# edd 21 Nov 2005
if [ -d /proc/hpc ]
then
/etc/init.d/portmap start
sleep 1
echo "Starting openMosix."
sleep 1
( cd /usr/share/knoppix-terminalopenmosixserver/ && \
cat templates/miniroot/etc/fstab >> /etc/fstab && cd - )
[ ! /mfs ] && mkdir /mfs
/etc/init.d/openmosix start
sleep 1
/etc/init.d/openmosixcollector start
fi
With this, openMosix started correctly on all my tests when I booted
"normally", i.e. a machine rebooted into the iso. But after the release of
0.7.9.1, I noticed that e.g. this code would not execute when the iso was
started with VMware Player. I suspect that may be due to /proc/hpc not being
there in an emulated/virtual session. Any additional light you could shed on
this would be appreciated!
| If there is no known cure, my plan is to edit the appropriate script(s)
| in /etc/init.d to check for and create /mfs if it doesn't already
| exist, and then remaster a local Quantian ISO and some DVDs. Should I
Keep in mind that we do not want this under kernel 2.6. So we do need a test,
but maybe a more robust one? Instead of testing for /proc/hpc, maybe testing
the uname output is better?
| feed this change (if I make it and have it work) back into
| clusterKnoppix, or just to Quantian (I presume there's interest here)?
Sure thing! Such a fix should definitely go into 0.7.9.2 as these are the
little things we are trying to get correct on incremental releases.
Regards, Dirk
--
Hell, there are no rules here - we're trying to accomplish something.
-- Thomas A. Edison
More information about the Quantian-general
mailing list