SCIENTIFIC-LINUX-USERS Archives

June 2011

SCIENTIFIC-LINUX-USERS@LISTSERV.FNAL.GOV

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
"Jeffrey R. Tunison" <[log in to unmask]>
Reply To:
Jeffrey R. Tunison
Date:
Fri, 10 Jun 2011 17:03:57 -0500
Content-Type:
text/plain
Parts/Attachments:
text/plain (61 lines)
Hi. Did the kickstart in SL6.1 stop mounting the installation tree on
/mnt/source?   I am just not finding the answer after googling and reading
and experimenting for 2hrs so I figure I'm doing something less bright. But
today I've come acrosss mention of more significant changes in RHEL6 than I
had thought from earlier reading, so maybe it's not listed in the RHEL 6
migration guide but is well-known by others?

I thought /mnt/source was a standard location during the kickstart to find
the installation media.

Today I was quickly trying to get kickstart configured with SL6.1a by
copying our successful SL5.4 kickstart files and modifying as necessary to
change paths. (And stop using the few deprecated terms in the kickstart
config file like mouse.)

I had previously divided up my ks files into various parts and then
%included them where necessary to try to generalize yet cope with our
various administrative domains needing different configurations, package
lists, etc.  It worked well for me. I believe I can work around my present
problem by creating single ks files, but I don't want to go backwards like that.

This is an NFS distribution of the package files.
First line of my ks file looks like:
nfs --server hidden.hostname.domain.edu --dir
/export/sysadmin1/kickstart/sl61/i386

PXE booting a 32 bit machine works fine with sl61 image. Anaconda runs and
then dies when it encounters references to /mnt/source in several places in
my kickstart file.

I see in the alternate consoles that the first line of my ks file is being
correctly parsed for the hostname and the directory to mount, but anaconda
fails with errors about being unable to access /mnt/source/blah mentioned in
lines like: %include /mnt/source/ks-sl61/i386.incl.domain
and then I see in the shell running in an alternate console that the only
NFS mounts are of /mnt/stage2 and /mnt/disk-image, neither of which contains
the rpm package files (they are deeper subdirectories of
/export/sysadmin1/kickstart/sl61/i386).  

So NFS and host resolution etc is all working but /mnt/source just seems to
be gone, with no errors that I can see in the alternate consoles.

Trying to work around this issue, I'm unsuccessful at specifying a %include
using an url, which I saw mentioned was now available in RHEL6. (I assumed
that it could be an NFS reference, maybe that was my flaw there...) None of
these forms have worked:
%include nfs:hostname:/path/to/file
%include hostname:/path/to/file
%include IP:/path/to/file
%include IP /path/to/file
%include nfs://IP/path/to/file

hmmm, that last effort failed with "PYCURL ERROR 1 - "Protocol nfs not
supported or disabled in libcurl."

Anyway, that's all fighting a battle which maybe I can avoid by knowing
whether /mnt/source is a standard that is failing to work for some reason or
is no longer available.

Thanks, and have a good weekend!

ATOM RSS1 RSS2