[37] in Moira
[haynes@UCSCC.UCSC.EDU: Scruffy notes, round 2]
daemon@ATHENA.MIT.EDU (Mark Rosenstein)
Thu Feb 15 11:02:26 1990
Date: Thu, 15 Feb 90 11:01:57 -0500
From: Mark Rosenstein <mar@MIT.EDU>
To: bug-moira@MIT.EDU
Date: Wed, 14 Feb 90 23:52:59 -0800
From: haynes@UCSCC.UCSC.EDU (99700000)
To: mar@ATHENA.MIT.EDU
Subject: Scruffy notes, round 2
First attempt to run the DCM. (I wish there was, maybe there is, an
easy way to make the dcm, for testing purposes, do all the data instead
of only the data that have changed since the last dcm run. And maybe
not really send the updates out to the servers, since I just want to
look at the results of the extractions.
gen/aliases.qc gen/mailhub.qc these append .LOCAL to a host name, which
our current sendmail.cf doesn't appreciate. So I'll change this to th
local domain name, or just leave it off and let the name server
figure things out from the bare host name
gen/aliases.sh has some MIT-isms in it for recognizing the mail hub
dcm complained about no /u1/sms/nfs directory, which apparently is needed
in addition to /u1/sms/dcm/nfs
hesiod: data are generated with UNSPECA I thought this had changed
to TXT Depends on what's happening with hesiod and BIND; but seems
like jb at Brown said he was going to make TXT part of the new BIND,
but was not going to include UNSPECA. Or something like that.
clusters: these came out
UCSCC.cluster HS CNAME smsinternal-1.cluster (comes from hesiod.qc)
I don't understand this, since I assigned the machines to a different
cluster name. Is there supposed to be an alias connecting the real
cluster name with this wired-in name?
service.db came out empty for some reason. I do have some services
defined and connected to machines.
Some other implicit lore, gradually being discovered:
MIT keeps prototype user dot files in /usr/prototype_user.
There are references to a setquota program and /etc/enforce-quota,
neither of which I'm up on.
I guess dcm expects the sms_update daemon to be running on all the servers
to be updated.
------
I made an unfortunate choice of the starting uid and gid when I created
the data base. Now I'm asking a local Ingres cognoscente for a one-liner
I can use to fix it, being illiterate in Ingres myself.
------
I remember now something was said about an empty libzephyr for those who
don't (yet) have zephyr. But at the time I forgot about this and just
removed -lzephyr from the makefile in the regtape directory and everything
compiled OK.
The regtape program no longer works, gives the error message
0..../regtape: Incorrect number of arguments
I guess this means the add_user query has changed?
The students program does work with the 'student.input' sample file
supplied, but then goes into a loop at the end doing queries get_finger_by_login
and get_user_by_name over and over. And in fact it blew the last entry,
putting the user in twice, once correctly and once with the first name
nulled out.
When I used the moira user menu to expunge the people who just got added
this way it did indeed expunge them, but reported "sorry, registered users
cannot be deleted" after each one. Does this mean something?
