I've just upgraded an experimental cluster running Coda to Red Hat 6.0.
I'm using kerberos5 for authentication with a PAM kerberos5 module.

I just got SSH to work correctly with kerberos and the PAM module so
that it creates a krb5 ticket cache file based on the process id, and then
deletes it when the user logs out.

I would like to be able to grant Coda tokens from the kerberos tickets on
login (this works already via kclog), and more importantly, *Delete* the
coda tokens for the corresponding process/tty when the user logs out.

I remember seeing something about Process Authentication Groups that
sounds like it might be what I want.. Could someone enlighten me as to how
to do this or point me at some documentation?

Related to this, I've got hesiod working for passwd/group distribution. I
would like figure out some way that Coda and hesiod could have coordinated
user and group data. Is this possible at all? Might it work to have a perl
script that takes some master data file and generates both the Coda users
and groups data files and the hesiod DNS records?

for some more ambitious ideas, what do we need to do to support Cells in
coda? (What about either making 'coda.foo.com' a TXT record with the
servers, or specifying that 'coda.foo.com' is round-robin record to all
the servers? Maybe even make 'scm.coda.foo.com' be the default SCM?)

Thanks

--------------------------------------------------------------------------
| Troy Benjegerdes    |       troy_at_microux.com     |    hozer_at_drgw.net   |
|    Unix is user friendly... You just have to be friendly to it first.  |
| This message composed with 100% free software.    http://www.gnu.org   |
--------------------------------------------------------------------------