[CUWiN] Soekris images not quite right?

[simon-cuw at uc.org]

I'm using the model with the onboard memory, and I used the Soekris 4526
specific image, not the sandisk images.

I was speaking with Jeff Brink, and he mentioned that he'd upgraded a couple of
his units using the upgrade script, and by doing that I got the desired
outcome:

   # /sbin/upgrade user at host:/path/to/upgrade.tgz

It creates a new filesystem on the spare /dev/wd0e partition and extracts a
tarball rather than writing an image directly. The resulting filesystem comes
out much better:

# df -k
Filesystem  1K-blocks      Used     Avail Capacity  Mounted on
/dev/wd0e       30239     22260      6468    77%    /
mfs:8              47        40         4    90%    /dev
mfs:1242         7223      1479      5383    21%    /mfs
/etc            30239     22260      6468    77%    /permanent/etc
/home           30239     22260      6468    77%    /permanent/home
/tmp            30239     22260      6468    77%    /permanent/tmp
/var            30239     22260      6468    77%    /permanent/var
/mfs/etc         7223      1479      5383    21%    /etc
/mfs/home        7223      1479      5383    21%    /home
/mfs/tmp         7223      1479      5383    21%    /tmp
/mfs/var         7223      1479      5383    21%    /var

In regards to rewriting configuration files -- for those of you who don't know
how to remount your partition for read/write, you do the following:

# mount -u -w /
# vi /permanent/etc/cuw_config
# mount -u -r /


Cheers,
Simon P. Ditner

On Wed, 1 Feb 2006, Bill Comisky wrote:

> On Wed, 1 Feb 2006, simon-cuw at uc.org wrote:
>
> > Hi,
> >
> > I decided to re-image a soekris node that was misbehaving (my first;
> > someone else had done all the previous configs), and I've run into a bit
> > of a problem...
> >
> > After installing the CUWiN 0.6.0 Soekris image using the metrix PXE boot
> > images, I noticed that the disk is not being fully utilized:
> >
> > # df -k
> > Filesystem  1K-blocks      Used     Avail Capacity  Mounted on
> > /dev/wd0a       22615     22010      -525   102%    /
> > mfs:8              47        40         4    90%    /dev
> > mfs:1242         7223      1456      5406    21%    /mfs
> > /etc            22615     22010      -525   102%    /permanent/etc
> > /home           22615     22010      -525   102%    /permanent/home
> > /tmp            22615     22010      -525   102%    /permanent/tmp
> > /var            22615     22010      -525   102%    /permanent/var
> > /mfs/etc         7223      1456      5406    21%    /etc
> > /mfs/home        7223      1456      5406    21%    /home
> > /mfs/tmp         7223      1456      5406    21%    /tmp
> > /mfs/var         7223      1456      5406    21%    /var
> >
> >
> > Disklabel shows that I should have more like 31248 blocks available:
> >
> > 5 partitions:
> > #        size    offset     fstype [fsize bsize cpg/sgs]
> > a:     62497        63     4.2BSD      0     0     0  # (Cyl.      0*-    488*)
> > c:     62497        63     unused      0     0        # (Cyl.      0*-    488*)
> > d:    125056         0     unused      0     0        # (Cyl.      0 -    976)
> > e:     62496     62560     4.2BSD      0     0     0  # (Cyl.    488*-    976)
> > #
> >
> > It seems to be causing me no end of trouble. The greatest of which is that if I
> > update anything in the web configuration, /etc/cuw_config gets wiped out, and
> > the node gets knocked off the network.
> >
> > Is there some trick to properly prep a soekris node that I've missed?
>
> What model Soekris are you running.. does it have built in CF memory or
> are you using a compact flash card?  If the latter it could be a problem
> of different disk geometry.
>
> I think the web configuration is misbehaving, as I've seen the same
> symptoms you have on a node with plenty of RAM/storage available.  You're
> better off for the moment mounting read/write, editing
> /permanent/etc/cuw_config, mounting read-only and rebooting.
>
> Bill
>
> --
> Bill Comisky
> bcomisky at pobox.com
>