File: ichain.txt

package info (click to toggle)
open-build-service 2.9.4-10
  • links: PTS, VCS
  • area: main
  • in suites: sid
  • size: 30,544 kB
  • sloc: ruby: 73,389; perl: 51,999; xml: 9,782; sh: 5,540; javascript: 2,743; sql: 1,232; python: 435; makefile: 229; cpp: 46
file content (104 lines) | stat: -rw-r--r-- 4,641 bytes parent folder | download | duplicates (3)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
Authentication on the openSUSE Build Service
============================================

The authentication used on the build system has the following
basic requirements:

* The authentication has to happen on the frontend since we need
  client independance. That means that a valid authentication check
  can only be done on the frontend. Clients only transfer credentials
  to the frontend.

* On the fronend, ActiveRBac for Ruby on Rails 
  ( https://activerbac.turingstudio.com/trac ) is used because we
  want to benefit especially from the permission system that ActiveRBac
  comes with. Every change to the openSUSE build service authentication
  system should not change the ActiveRBAC code. Moreover our changes
  should integrate into the ActiveRBAC concepts.

* Additionally we want to integrate the iChain support with the ActiveRBAC  
  user management. iChain is a Novell identity management solution 
  and if we support users can do everything with one login on the
  Novell and openSUSE websites.

* Since there are many registered users on the Novell websites we 
  do not want to give access to the openSUSE build service by default
  to anybody. Especially in the beginning we want to be able to
  control who has access to the BS. Thus there needs to be a process
  where people ask for enabling their iChain account and the BS admin
  team activates the user.

* iChain works like a proxy transparent in front of a web application.
  All it does is adding a header value to the http header. In our case
  (iChain is much more powerfull than this) we only have the username 
  in the header. Note that as a result of the administrational setup 
  this header value can be considered as 100% true and secure. That 
  means if the header contains the user name 'freitag' it is absolutely
  secure to consider that the user freitag is logged in. The huge benefit
  from that is that the app (the BS) does not have to bother with sensitive
  information like passwords for example.

* We want to control in the clients what pages or functions are accessible
  without authentication (ie. the start page) and which need an authenticated
  user. That should not need configuration in the iChain system.

How does that work in the BS?
=============================

Every controller on the webclient calls the extract_user method in
the application controller. This method tries to extract credentials
from the request depending on the kind of authentication that is 
running. 

To switch on ichain authentication the parameter ICHAIN_HOST in the 
config file on both webclient and frontend need to be set to the IP
address of the ichain host (Note: The IP is not yet used, so setting
to non-nil is sufficient).

If iChain is running and the user accesses a page that should be 
authenticated but is not yet, the webapp redirects to a special page.
The iChain system is configured in the way that accessing this special
page requires authentication and thus iChain displays the Novell
standard login page. The user provides the credentials and after the
login was successfull, the user is redirected to his initial requested
page.


If iChain is active the user is taken from the header value X-username
(HTTP_X_USERNAME) that iChain transparently adds to the HTTP header.
The username is added as a header value to all communication to the
frontend as well. 

To check if a user is valid, the webclient does a lookup on the user 
name on the frontend. 
That can result in the following states:

* the user was found and has the ActiveRBAC state 2: The user is valid
  and allowed to login in. 

* the user is not found. That means that the access to the BS was not
  yet granted. The user is forwarded to a page that lets him ask for
  BS access. 

* the user was found but is in state 5: The user has already asked for
  BS access but the BS admin team has not approved. The user sees a 
  message that asks him to wait.

In case the user is not yet in the frontend user database, the client
sends a XML document of the following form to the frontend controller
person action register:

       <unregisteredperson>
           <login>freitag</login>
           <realname>Klaas Freitag</realname>
           <email>freitag@suse.de</email>
           <state>5</state>
           <password>opensuse</password>
           <note>This is why I like to work with the BS</note>
        </unregisteredperson>

This controller adds the new user to the BS database and sets the
state to be unconfirmned. Now a BS admin has to switch the users
state to confirmned. As long as that has not happened the user can
not log in correctly.