No Subject

[Anonymous]

> > No, I'm just doing a klog.  Formerly I had klog as a script that
> > just did kauth (or kinit; I forget which), then afslog.  Now I'm
> > using the new klog from arla, and it doesn't get Kerberos tickets.
> 
> The klog seams to just keep the tickets if you give it the option
> -tmp. That might change.
> 
> > Should it?  If it doesn't, then shouldn't the other arla commands
> > work by cheking for tokens rather than tickets?  Otherwise, shouldn't
> > the arla klog get kerberos tickets as well as tokens?
> 
> I think it should get tickets and tokens. The tools in arla should check
> for both tickets and tokens (in that order). We want to keep compability
> with the transarc tools. The general idea is to be better.

Transarc's klog only gets a Kerberos ticket if -tmp is specified on the
command line. (I think that that is a horrible choice of name, because you
shouldn't put Kerberos tickets in /tmp to begin with)
To be compatibility with the Transarc tools we really need to keep it this
way. It would be nice to be better, but we don't want to blindly break all
existing scripts and confuse users who are used to the Transarc behavior.

One thing to keep in mind is that if klog gets a ticket, it will overwrite
any existing ticket; if you want to get a token for another cell, you
don't necessarily want to destroy an existing Kerberos ticket.


- Chris Wing
wingc at engin.umich.edu

"author of the stupid klog program" :)