Hi, On Sun, Apr 01, 2007 at 09:56:24AM -0700, Adam Megacz wrote:
[snip] > I see. I think nscd would achieve the same effect... Not really. It would still mean, having to wait for missing DB-servers to be marked offline - for each process. > Also, while I'm hacking on this, two more questions: > > 1. Does it make sense to map AFS group ids (with the sign reversed) > onto GIDs? Most likely no. > 2. Previous suggestions about shells involved putting users into > groups named after shells, such as > > pts add adam shell:bash > > But it seems that this wouldn't allow users to manage their own > shells since shell:bash would belong to some administrative user. > What about an alternative, where users would create a group > "username:shell.bash"? > > pts cg adam:shell.bash Hmm... Where does the 'groupname'->'shell-binary' map come from? When I started using AFS I despaired of Kerberos. My boss suggested, simply to put a passwd-file into the AFS-filespace :-) . Not a really good idea for authentication but maybe for a user-shell-map. Changing the shell could be accomplished by some web-interface and a script, updating this file once an hour. > Davor Ocelic also had a neat suggestion about giving human-readable > names to PAG groups. Do you have some pointer for that? Regards, Frank _______________________________________________ OpenAFS-info mailing list [email protected] https://lists.openafs.org/mailman/listinfo/openafs-info
