bernerus@cs.chalmers.se (Christer Bernerus) (02/22/91)
I'm currently working on a project of installing an AMDS system based on
ATK PL9, AFS3.1 and RT_AOS.
Below are some problems I have encountered during the process of
installing AMDS and my efforts of getting netnews into thei system:
1. To get nntppoll going, one must initialize a file called
last_time. This wasn't done by pobbconf and I found no
documentation telling me what to put in there. Looking into the
NNTP RFC document and the sources gave me the following format:
a_numeric_code<whitespace>YYMMDD HHMMSS.
Initializing this file with -1 870101 000000 made nntppoll going.
2. Now that nntppoll is working, it seems to put zillions of
files in /usr/net/Hold/* and eventually also in
/usr/net/ReadyBox. As I understand it, then nns is expected to
look in ReadyBox and move the articles into its final position
in ../netbb/.MESSAGES/*. However, there seems to be no provision
made for creating the destination folders when nns is running,
and the articles that have no corresponding "newsgroup" folder
won't be copied. Do I have to go around and create all newsgroup
folders by hand? and what happens later when new newsgroups are
created ?
3. The file rc.local created by pobbconf assumes that there is
an /bin/su that can take a password on the command line. I
didn't find any, especially not one that authenticates to AFS.
4. Fortunately on my system root doesen't need a password for
su'ing to someone else, so it was relatively simple to write a
csh script that does a su postman, then klog -pr postman -pa
password and then executes /etc/startup. Now, my sendmail
(oldsendmail) will refuse to execute, since there seems to be
some paranoid uid check somewhere in sendmail. Is it safe to run
sendmail before su'ing to postman, and do I need to have
postman's AFS tokens for (old)sendmail ?
5. I'd wish that the sample configuration of AMDS (pobbconf.c)
had the CMU local setups removed. This file is almost impossible
to understand, and the configuration is really a nightmare at
least when doing this the first time. Now that things seem to be
working, I don't have the faintest idea if what I've done is
really what I wanted.
6. The configuration process uses package. While being an
invaluable tool for administering the local disks on the
workstations, I lack one important option in package: A report
mode. Debugging the configuration is now an extremely tedious
task since there is no way to produce a report on all the
inconsistencies in the configuration files, the "source" tree
and the local disk setup. For almost every error, package stops,
and won't let me see the rest of my errors until I've fixed the
first one found. This is very irritating. I don't know who's
responsible for package (Transarc ?), but I hope they'll take
note.
There have been lots of other problems as well, but those above are the
most significant ones. The AMDS system seem quite promising once
installed, but I'd be happy for comments and help from those of you who
know whats going on inside this and also from anyone else who have gone
through the process of installing AMDS.
Chris.
-------------------------------------------------------
Christer Bernerus ! E-mail: bernerus@cs.chalmers.se
Chalmers University of Technology ! Phone: +46 31 721000
Department of Computer Science ! Ham radio: SM6FBQ 144.3 MHz
S-412 96 Gothenburg, SWEDEN