Coda File System

Re: Automated clog in login scripts

From: Jan Harkes <>
Date: Tue, 24 Feb 2004 12:16:08 -0500
On Tue, Feb 24, 2004 at 10:34:25AM -0600, Jason A. Pattie wrote:
> | Hmm, that does complicate things a bit, because disconnection/connection
> | is really on a per-volume basis, and as such we don't really know
> | whether we are connected until we try to access the servers.
> |
> | Maybe add some kind of `(f)ping -c 5 servername` test?
> What about 'cfs checkservers'? or similar?

Not too useful. cfs checkservers really only checks if we can reach the
servers that host volumes that we currently have cached.

So when I start a client without doing anything, it will always return
success because it thinks we do not need any servers. If I 'ls
/coda/' it will return success whenever verdi, mozart and
marais are reachable. Which is again not all that useful for me, because
I also need vivaldi and mahler before I can access my Coda home

And if I happen to access /coda/<some other remote realm> at some point
and those servers happen to be unreachable, cfs checkservers will fail.
Even when I have perfect access to all the servers that I need to get my

In this case you really are concerned whether you can reach the auth2
servers. These can be totally different from the coda volume servers, so
the real application to tell us about them would be clog.

Right now it doesn't really distinguish between typing an incorrect
password and not being able to reach the servers, but it could exit with
different error codes depending on the type of problem.

The clog part would become something like,

    clog -tofile foo @realm
    if [ $? -eq 2 ] ; then
	clog -fromfile foo @realm

(ofcourse there isn't such a useful exit code at the moment).

Received on 2004-02-24 12:18:00