Hi Steve and Ming-Hong,
could you please try to get some more info in these boxes?
I am afraid that German has no chance to find out what the problem
is if he doesn't get access to a similar box.
Chip-set, disk (size and type) would be a start.
If you can't use older boxes to make progress it might be
useful to have a look at the bios setting of these boxes.
If you spot some advanced feature related to disk I/O, switch it
off and retry.
Until we get some more information you can try the following things:
1) Go to the configuration file in lcg1/source/redhat73-cfg.h
look there for the fstab commands.
In the past we had on some nodes problems wit the
stab.mkmpt_hdaX zap
option.
Replace the "zap" by "ignore"
I don't have much confidence in this, but it will be a quite
quick test.
Redo the profile and reinstall the node. Before you do
use a rescue disk and make sure the disk has no more partitions
2) If you are still reading I'll assume that 1) didn't work, too bad.
If you use a large disk try to limit the size used by LCFGng.
We don't know if there is a problem, but we never used the given
config on disks larger than 40GB.
In the same file as above go to the line
fstab.size_hda2 free
and replace free by the size in MB that you want the / partition to be.
Be, for the test, conservative and set it to 15GB (1500).
Same procedure as before, clean, set the zap back, make the profile,
try an install.
3) Still reading.....
Try 2 with replacing the zap by ignore
4) Now we will go into desperation mode.
Take your rescue disk and partition the disk and make a filesystem
following the parameters that you used in the redhat73-cfg.h file.
Now go to the file an switch the option :
fstab.preserve_hdaX no
from no to yes. This should keep lcfg from rebuilding the
file system (never tried this)
Install the node
5) Toggle the zap and ignore and do step 4 again.
No more ideas what you could try.
6) Find someone who is willing to temporary give you old boxes
in exchange of your new boxes
Good luck
markus
On Mon, 21 Jul 2003, Steve Traylen wrote:
> > > [INFO] updaterpms: installing filesystem-2.1.6-2
> > >
> > > [WARNING] updaterpms: unpacking of archive failed on file /boot: cpio: chown failed - Read-only file system
> > > ......
>
>
> Ming-Hong,
>
> I have just seen exactly this error here on a box that some one was
> installing here. This is also a very new box and is the first time
> I have seen this error. No ideas at the moment.
>
> Steve
>
>
> >
> On Mon, 21 Jul 2003, Markus SCHULZ wrote:
>
> > On Mon, 21 Jul 2003, Ming-Hong Bai wrote:
> >
> >
> > > Dear Markus,
> > >
> > > I am one of the administrators of Taipei Site.
> > > We have met a problem when install the LCG-1 testbed.
> > > When I install each element, I got some warning messages like following:
> > > -------------------------------------------------------------------------------------
> > >
> > > [INFO] updaterpms: installing filesystem-2.1.6-2
> > >
> > > [WARNING] updaterpms: unpacking of archive failed on file /boot: cpio: chown failed - Read-only file system
> > > ......
> > >
> > > INFO] updaterpms: installing grub-0.91-4
> > > [WARNING] updaterpms: unpacking of archive failed on file /boot/grub: cpio: mkdir failed - Read-only file system
> > >
>
>
>
> > > ......
> > >
> > > ---------------------------------------------------------------------------------------
> > > And I checked the partitions of /dev/hda, found that the partition table
> > > seems not correct.
> > > The total blocks number shown by 'df' command is twice larger than the
> > > hard disk has, and
> > > the /boot partition has almost 80G space. The /boot (/dev/hda1)
> > > partition can't be write,
> > > because the partition is strange. Every node in our testbed have the
> > > same problem. But
> > > when I use an older machine to install , it can be installed
> > > successfully. I am not sure if the
> > > partition tools of LCFGng have bug for some new hard disks.
> > > Do you have any idea?
> > >
> > > Ming-Hong Bai
> > >
> > > >
> > > >Site: Taipei
> > > >Checkout of the initial release from cvs: yes
> > > >Requested write access to my sites module: no
> > > >Got the write access: no
> > > >Checked in first changes: yes
> > > >Have a working LCFGng server: yes
> > > >Have started to install my nodes: yes
> > > >Have finished the following nodes: no
> > > >The following nodes are missing: none
> > > >
> > > >Have setup the site status page: yes
> > > >I have updated the
> > > >site status page (since the 16th): no reason
> > > >I have registered or received a
> > > >registration mail for [log in to unmask]: yes
> > > >I have signed on to the IM: no
> > > >I have registered with dteam VO: no
> > > >
> > > >I have problems with the following components: all
> > > >I can't participate in the current deployment
> > > >activities for the following reasons:
> > > >(something like: no time, no nodes, don't see why I should, no power...)
> > > >
> > > >
> > > >
> > > >Again I put the generic site names and individuals in the CC.
> > > >I got only confiramtion from lyon that their generic site list is
> > > >monitored.
> > > >
> > > > markus "form" schulz
> > > >
> > > >
> > > >
> > > >
> > > >--
> > > >*************************************************************************
> > > >* *
> > > >* CERN Markus W. Schulz *
> > > >* Bat. 31 2-015 *
> > > >* CH-1211 Geneva 23 *
> > > >* *
> > > >* Phone: +41 22 76 77909 *
> > > >* www.cern.ch *
> > > >* *
> > > >*************************************************************************
> > > >
> > > >
> > > >
> > >
> > >
> > >
> >
> > --
> > *************************************************************************
> > * *
> > * CERN Markus W. Schulz *
> > * Bat. 31 2-015 *
> > * CH-1211 Geneva 23 *
> > * *
> > * Phone: +41 22 76 77909 *
> > * www.cern.ch *
> > * *
> > *************************************************************************
> >
>
> --
> Steve Traylen
> [log in to unmask]
> http://www.gridpp.ac.uk/
>
--
*************************************************************************
* *
* CERN Markus W. Schulz *
* Bat. 31 2-015 *
* CH-1211 Geneva 23 *
* *
* Phone: +41 22 76 77909 *
* www.cern.ch *
* *
*************************************************************************
|