Jason... Seem to keep running into you on various mailing lists!! Thanks for the input.. I wasn't aware that RH linux had implemented ACLs (ala Windows)... I was aware of the chroot/jail concept.. And yes, you've nailed the issue that we might face which is how to deal with the "truly" ruthless.... So we're trying to consider a number of possibilities as solutions to this issue.. a possible solution is to allow text development on one server... restrict the user from running anything except some basic apps. chroot/jail the user as well.... allow the user to do compiles/builds on another server, with the idea being that the "build/compile" server would essentially pull the user's code from a CVS app, and then automatically build the code.. this approach would allow code to be built.. but this is ugly, and imposes restrictions that we're not sure is really necessary... a basic question we have... if we effectively chroot a user, what are the real issues involved with letting that user then build/compile apps under this space... I'm assuming that there are ways to restrict users from being able to run given apps. we'd want to let the user build/compile the app but not to run the app on the "build" machine... also, are there ways to essentially lock down a given machine, such that you can run an application on the machine, but the application is restricted to either its' own process/resource space or that it couldn't do any damage beyond the given machine that it's being run on.. I ask because we could always have "test" servers that are essentially rebuilt after every application .bin is tested... We really need a seriously good security guru to bounce these thought off of!!!! Thanks... -Bruce bedouglas@xxxxxxxxxxxxx -----Original Message----- From: redhat-list-admin@xxxxxxxxxx [mailto:redhat-list-admin@xxxxxxxxxx]On Behalf Of Jason Dixon Sent: Friday, December 05, 2003 4:31 AM To: Red Hat Mailing List Subject: RE: Security Issues.... On Fri, 2003-12-05 at 03:52, bruce wrote: > Ed... > > You're getting it!! But given that the enviornment that I envision will have > literally strangers coming into my box... I need to be reasonable > paranoid/secure... Which brings me to my original question... I think I've > pretty well formed the basis of the issue/problem that I'm seeking to > solve... I don't envy your situation. If you're going to give folks carte blanche with compilers and development tools, it's only a matter of time before the truly ruthless break your chain of "control". You need to develop a full-blown security policy. Some concepts to Google: - ACLs - chroots/jails - sandboxing (rebuild environment at regular intervals) - Intrustion Detection (Host and Network) - Social engineering HTH. -- Jason Dixon, RHCE DixonGroup Consulting http://www.dixongroup.net -- redhat-list mailing list unsubscribe mailto:redhat-list-request@xxxxxxxxxx?subject=unsubscribe https://www.redhat.com/mailman/listinfo/redhat-list -- redhat-list mailing list unsubscribe mailto:redhat-list-request@xxxxxxxxxx?subject=unsubscribe https://www.redhat.com/mailman/listinfo/redhat-list
