[OpenAFS] Re: AFS and freedesktop.org trash

omalleys@msu.edu omalleys@msu.edu
Wed, 01 Jun 2011 13:57:55 -0400

Quoting Andrew Deason <adeason@sinenomine.net>:

> On Wed, 1 Jun 2011 10:51:42 -0600
> Ken Dreyer <ktdreyer@ktdreyer.com> wrote:
>> I was curious if anyone has experience with AFS and the
>> freedesktop.org "trashcan" spec.
>> http://www.freedesktop.org/wiki/Specifications/trash-spec
>> The spec references a "topdir" area where a ".Trash" or a
>> ".Trash-$uid" directory can exist at the top of a filesystem mount.
>> For AFS, that translates to /afs/.Trash . With -afsdb enabled, this
>> means doing several DNS lookups for a "Trash" AFS cell.
> Don't we do some kind of blacklisting on OS X for things like this? We
> could just ignore dynroot requests for .Trash. It's a legit name to have
> there, though.
> But also, shouldn't negative DNS responses be cached for awhile? I
> wouldn't think each of these accesses should entail a long delay.

It depends on how your system is configured. If you are using  
something like nscd, then you can adjust negative ttl's in the  
nscd.conf file. dnsmasq uses the ttl handed out by the dns server in  
the SOA record which bind9 is clarifying that as the minimum ttl.  If  
none is set then it is 0. (There is an option to override this  

dnsmasq also doesn't doesnt cache tcp dns requests at all. (according  
to their faq.)