[1319] in athena10
Re: [Debathena] #124: De-crustify the CellServDB
daemon@ATHENA.MIT.EDU (Jonathon Weiss)
Fri Mar 6 13:48:11 2009
Message-Id: <200903061844.n26Iiast010569@wax-lion.mit.edu>
From: Jonathon Weiss <jweiss@MIT.EDU>
To: Evan Broder <broder@MIT.EDU>
cc: Jonathon Weiss <jweiss@MIT.EDU>, debathena@MIT.EDU
In-reply-to: Your message of "Fri, 06 Mar 2009 13:19:20 EST."
<49B16928.5070304@mit.edu>
Date: Fri, 06 Mar 2009 13:44:36 -0500
> Jonathon Weiss wrote:
> >> Since this apparently wasn't clear, I realize that ops maintains the
> >> CellServDB, but its crustiness creates a bad user experience for anyone
> >> who tries to access other cells, so I want to take responsibility for
> >> figuring out what the necessary cleanups are.
> >>
> >
> > We update ours from /afs/grand.central.org/service/CellServDB
> > occasionally. If there are broken things in that, you can try bugging
> > jhutz at cs.cmu. If the problems are that we are out of date from
> > that, then a note to afsreq may reduce the delay to the next update.
> >
> > Jonathon
> >
>
> Would you be open to trimming down the CellServDB to a shorter list of
> cells than the full list maintained in grand.central.org? Since all
> current Athena systems can use DNS these days (they can, right?), it
> seems like having entries for cells that aren't essential to Athena is
> just asking for them to become out of date.
>
> - Evan
I'd need to discuss this issue with other folks here befor I can
answer that one way or another. There are two issues:
1) I'm not sure we want to be in the business of maintianing a
set of diffs from the cannonical CSDB for not-MIT cells
(though I did drop jhutz a note asking about the idea of a
seperate CSDB for clients that have afsdb and dynroot.
2) some (really old) clients don't support afsdb or don't have
it enabled by default. We'd need to decide if we're
comfortable with them losing access though the relevant
foreign cells. Actually, I recently came accross some new
clients that didn't have afsdb enabled by default (though
enabling it was realtively easy), so that may also be a
concern.
What are the negative user experiences that you are referring to?
With the fakestat option turned on, I think 80-90% of the problems go
away.
Jonathon
