[OpenAFS-port-darwin] configuring openafs under MacOS 10

[Ted McCabe]

At 11:59 AM -0700 6/11/01, Shantonu Sen wrote:
>>  I would do something along these lines.  Have a configuration
>>  setting which says *which* cells should be added at system
>>  startup, and then something else which can be used to parse
>>  the "CellServDB.full" file and execute 'fs newcell' commands
>>  to bring up cells on the fly as the user requests them.
>
>So cells could be marked as boot-time-active or not, and those cells 
>would get written out to CellServDB. Additionally, cells can be 
>activated dynamically, but they wouldn't be started on next boot?
>
>This seems reasonable. Would we use another file to track this? I 
>don't like the thought of checking each cell to see if it has been 
>activated, so I'd prefer the have the info written to a file.
>
>Shantonu

For a central admin, keeping a CellServDB (used at boot time) and 
CellServDB.full (used for reference) is likely the easiest route, 
plus any auxiliary files to keep the scripts/tools from being too 
complex.

First question: Is this going to be a two-part tool, i.e. a GUI 
wrapper for a command-line prog?  If so, then the command-line 
prog/script could be used by admins of clients that aren't Macs.

Second question: any thoughts as to how to manage this when AFSDB RR 
comes out?  I'd hate for an interface that prunes the CellServDB to 
be developed only to become meaningless if the Finder can still look 
up cell info over DNS.

A related question:  this list maintenance is similar to Athena's 
config_afs script, which has both the site's preferred 
CellServDB.public and a client's local additions in CellServDB.local. 
It also manages the list of which cells he client can trust suid. 
Will this tool behave similarly?

    --Ted