[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: ACLs



Quoting sean finney (seanius@seanius.net):

> assuming you mean exclusively filesystem acls, acl support is
> there... kind of.  for ext2/ext3, there's a kernel-patch package
> you can use to build your own kernel quite easily the Debian Way
> with make-kpkg.
> 
> the trouble is stuff you already hinted at, and a little more. for ext2
> and ext3 filesystems:
> 	- they don't work with quotas
> 	- they don't work with stock linux nfs (is there a nfs patch?)
> 	- they also don't work on smp machines at all
> 	- they are only partially supported by many of the userland utils

(I assume you're talking about the POSIX ACLs implementation at 
http://acl.bestbits.at/ .  Following is from Andreas Grünbacher's
write-up at http://acl.bestbits.at/problems.html .)

ext2 doesn't seem to have been a problem; ext3 for a long time _had_ a
problem with machines where ACLs/extended attributes were in use, _with_
quotas, _with_ SMP support -- but that's now believed to be fixed.

Linux NFSv2 code doesn't support the ACCESS call to set/retrieve ACLs,
but there's a patch for the Linux NFSv3 client, as well as for the two
server-end alternatives.  The bigger problem for heterogeneous sites is
that there's no agreed-upon standard in the NFS definition for exactly
how ACCESS is to be implemented, so such tend to be vendor-specific.

-- 
<BLINK>Resize your browser so the following line touches both margins!</BLINK>
                           <HR WIDTH="75%">
                Best Regards, Rick Moen, rick@linuxmafia.com



Reply to: