It seems more effort has been put into arguing why not to do a simple 
RBAC profile for these command than it would have actually taken do it it.

If the project team doesn't actually know what privilege the commands 
need to run then I'd say they architecture isn't ready for review.  Just 
saying that they all might need to run with euid=0 isn't good enough in 
my opinion - particularly given how dangerous some of these commands can be.

We aren't asking for anything really complex we aren't even asking for a 
single line of code change.

Given that all of these commands need to actually be tested to work 
before integration all that needs to be done is work out what privileges 
they need when their functionality testing is being done.   The ppriv 
command with -D should be enough here.   I believe the project team has 
already done some if not all of this.

On the other hand I think Jim and Casper may have a good point here that 
because these commands are so possibility dangerous we wouldn't actually 
want to hand them out to some one via an RBAC profile because what can 
be done with them could be essentially equivalent to handing out all 
privs and uid=0 anyway.  If that is the case then I'll agree no RBAC 
profile for these is necessary - but might still be nice to have.

I'm not going to hold up this cases any further.  I'm getting to the 
point I think that ARC review is useless given all the push back we get 
from project teams about doing anything with respect to the ARC best 
practices.

--
Darren J Moffat


Reply via email to