1 |
On Tue, Aug 10, 2010 at 2:50 PM, Alan McKinnon <alan.mckinnon@×××××.com>wrote: |
2 |
|
3 |
> On Tuesday 10 August 2010 15:03:19 Kevin O'Gorman wrote: |
4 |
> > On Mon, Aug 9, 2010 at 6:18 PM, William Hubbs <williamh@g.o> |
5 |
> wrote: |
6 |
> > > On Mon, Aug 09, 2010 at 05:30:40PM -0700, Kevin O'Gorman wrote: |
7 |
> > > > On Mon, Aug 9, 2010 at 1:20 PM, Bill Longman <bill.longman@×××××.com |
8 |
> > |
9 |
> > > |
10 |
> > > wrote: |
11 |
> > > > > I actually prefer "sudo su -" -- as long as I'm giving it away! |
12 |
> :o) |
13 |
> > > |
14 |
> > > Afaik, there is no reason for "sudo su -" It should be either |
15 |
> > > |
16 |
> > > su - |
17 |
> > > |
18 |
> > > or, if you are using sudo, |
19 |
> > > |
20 |
> > > sudo -i |
21 |
> > > |
22 |
> > > The disadvantage of "su -" is that it requires the user to know the |
23 |
> root |
24 |
> > > password. But, "sudo -i" does the same thing without requiring the |
25 |
> user |
26 |
> > > to know the root password. |
27 |
> > > |
28 |
> > > You either didn't think or didn't actually try it. "sudo su -" needs |
29 |
> a |
30 |
> > |
31 |
> > password, but it's the |
32 |
> > user password. Running su as root never needs a password. Accordingly, |
33 |
> > this works on |
34 |
> > a stock Ubuntu with no root password. |
35 |
> > |
36 |
> > "su -" requires the root password unless you're already root, and the |
37 |
> root |
38 |
> > password may or may not exist. |
39 |
> > |
40 |
> > I didn't know about "sudo -i" (thanks), but when I tried "sudo -i" it |
41 |
> > immediately asked for a password, for which |
42 |
> > the user password was sufficient. So it's entirely equivalent to but |
43 |
> > slightly shorter than my version. I'll stick with |
44 |
> > mine because it's made of parts I already know and won't forget. |
45 |
> > |
46 |
> > I think that if sudoers don't need to enter passwords, they're still |
47 |
> > equivalent, but I have not tried this. |
48 |
> |
49 |
> Sounds to me like he's whinging about sudo and not much else. I find this |
50 |
> to |
51 |
> be common and far too many people advancing the idea can't define to me |
52 |
> basic |
53 |
> security concepts. I have also yet to meet someone with a beef against sudo |
54 |
> that can show a fundamental weakness with it, and I'm not talking about an |
55 |
> isolated case of buffer overflow either - that can happen with any |
56 |
> software. I |
57 |
> mean a weakness in the methodology of sudo itself. |
58 |
> |
59 |
> Many people have a stuck idea in their heads that the root password is a |
60 |
> magic |
61 |
> security bullet. In fact, it's no such thing. Like any other password it is |
62 |
> simply something you need to prove you know in order to to authenticate |
63 |
> yourself. The major threat by analysis on a workstation is stepping away |
64 |
> for a |
65 |
> leak and forgetting to lock the screen. sudo is adequate protection against |
66 |
> this as long as more than 5 minutes have elapsed since the last sudo was |
67 |
> run - |
68 |
> the prankster may have access to the machine but still does not know any |
69 |
> password, including yours. A major threat to finding passwords is shoulder |
70 |
> surfing. If one frequently enters the root password, it is equally easy for |
71 |
> a |
72 |
> shoulder surfer to find it as to find the user's password. Note that if you |
73 |
> leave your workstation unlocked with a root session open, there is no such |
74 |
> timeout as what one has with sudo. |
75 |
> |
76 |
> Additionally, on a shared machine (i.e. server at work), the root password |
77 |
> has |
78 |
> to be shared which is a huge hole in itself due to the difficulty of |
79 |
> communicating the new password when it is changed. It is trivially easy to |
80 |
> communicate a single password for a single user and guarantee it stays |
81 |
> secure |
82 |
> (major advances in cryptanalysis excepted). |
83 |
> |
84 |
> |
85 |
> -- |
86 |
> alan dot mckinnon at gmail dot com |
87 |
> |
88 |
> Good Luck getting people to change them frequently and haveing your techs |
89 |
and it departments meeting complexity and length policy. |
90 |
|
91 |
Remeber the only secure system is off and disconnected. |
92 |
|
93 |
If you are willing to use it you must apriase the community of the risk of |
94 |
failure; and plan for said risk. |
95 |
|
96 |
Most projects I've enjoyed had various password books usually encrypted with |
97 |
a "God" key for each department and it's respective responsbile area. |
98 |
|
99 |
Then those keys become an issue in and of themselfs; then it's a matter of |
100 |
procedural control. When the admin or admins leave, change them. |
101 |
|
102 |
Sounds simple, but far too rarely as it happens in pratice that I've headed |
103 |
to a client I haven't visited in a decade or so and find the same password I |
104 |
once used by guessing. |
105 |
|
106 |
Wich always rings true for me as a means to ensure disclosure is to those |
107 |
that I trust; or would trust. |
108 |
|
109 |
The discretionary access model in Gentoo is nice and to be expected; what |
110 |
I'd really like is a way to have my groups integrate from whichever |
111 |
directory service I'm using to meet the DAC mappings required on the local |
112 |
machine so I can enable RBAC or some other Lattice based control with local |
113 |
admins and limit their functions to thier jobs in an EASY fashon. |
114 |
|
115 |
Regards, |
116 |
-- |
117 |
Hazen Valliant-Saunders |