[986] in Moira
moiradev vs. opssrc vs. the server
daemon@ATHENA.MIT.EDU (Dan Winship)
Sun Jul 14 15:39:09 1996
Date: Sun, 14 Jul 1996 15:38:53 -0400 (EDT)
From: danw@MIT.EDU (Dan Winship)
To: moiradev@MIT.EDU, moira-admin@MIT.EDU
(Message 2 of N in a series on moira sources :-)
The moira sources are now in a CVS repository in moiradev. The main
branch of the CVS tree is what was in moiradev/src before (mostly... a
few changes were needed since you can't have symlinks in a CVS tree,
and I also imported a few fixes from opssrc). There are also branches
called OPSSRC and MOIRA4.
The OPSSRC branch is not quite what was in ~opssrc/moira/src before,
because that wasn't quite what was running on the servers. We grepped
through the server binaries for RCS ID strings, and asked a few people
which copies of the sources they had hacked on last, and tested the
behavior of the server and the DCMs, and so now I'm pretty sure that
the OPSSRC branch contains almost (but not quite exactly) what's
running on moira. I built the binaries and installed them on SOS and
did an mrrestore and DCMs, and only noticed a few differences in the
outputs:
- hesiod.gen on moira2 generates an entry in services.db for
clh.service HS UNSPECA "clh tcp 7971"
even though there's no `clh' entry in the services table... we're
not quite sure where it's coming from, but the hesiod.gen I built
and installed on SOS _didn't_ generate this entry. (Suspiciously,
clh is the last entry in the table [before the aliases], so it
might be an off-by-one error somewhere? It doesn't appear in the
backups either, though.)
- the passwd.db from moira2 has a few entries with odd spacing that
the hesiod.gen from sos `fixed':
< jliaw.passwd HS UNSPECA "jliaw:*:4000:101:Joey Liaw, Relief,Goddess Office,,56325:/mit/jliaw:/bin/athena/tcsh"
---
> jliaw.passwd HS UNSPECA "jliaw:*:4000:101:Joey Liaw,Relief,Goddess Office,,56325:/mit/jliaw:/bin/athena/tcsh"
- pobox.gen on moira2 seems to have had the patch from [887] in the
moira discuss meeting applied to it, but that patch wasn't applied
to opssrc or moiradev. (This has to do with fixing entries that
appeared in mitdir with no `@mit.edu' after the address.)
- nfs.gen didn't generate all of the same files on moira2 and sos,
but it copies some things from /moira/nfs, and sos didn't have that
directory, so it's probably only those files that were different.
Also, afs.incr doesn't build currently, because the afsdev locker has
been rearranged. (I tried changing the path to something else that
looked right, but the libraries there gave link errors.) And while
we're talking about afs.incr, does anyone know why that directory
isn't world-readable like the rest of the tree? (There don't seem to
be any bits of AFS source in it.)
I have a copy of OPSSRC checked out in ~moiradev/cvs-src/ops right
now, but it probably won't stay there... I'll get Jonathon to do
something permanent with it. Once that happens, the moira-admin and
moiradev people probably ought to figure out a better way of keeping
the trees in sync with each other and with what's running on the
server.
-- Dan
