Hi Martin,
You will have to expire/invalidate the session in the code upon user logout.
This way when the cookie comes in, there is no corresponding session-ID and the
system will create a new session. Are you doing that already? Does that help?
-Shanti
On Sep 20, 2011, at 1:20 PM, Christopher Schultz wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Martin,
>
> On 9/18/2011 11:05 AM, Martin O'Shea wrote:
>> I have a situation where I'm using Tomcat 6.0.26 but the logging in
>> / out of the application is not authenticated via Tomcat's:
>>
>> action='<%= response.encodeURL("j_security_check") %>' >
>>
>> method.
>
> You mean to say that you are using your own authentication mechanism,
> right?
>
>> The current system allows cookies to store userids which are used
>> to show recent lists on the homepage of the application. So for a
>> session, a user's userid can be read from the cookie and used to
>> retrieve their details from the database and store them in the
>> session, and render the hompage with its personalised recent list.
>
> So, any remote user can provide a forged cookie to read anyone's
> "recent list" if they want? You might want to encrypt those cookies.
>
>> The user's id can also then be placed in the login username box
>> with the password stored in the session.
>
> So, you use an untrusted user id coming from a remote cookie to
> populate the user's username and password on a login page? Sounds like
> that's a problem.
>
>> But, in a single browser session, if the first user logs out, and
>> another user logs in, the cookie is re-written with the new user's
>> userid. But, because this is all in one browser session, use of the
>> browser's back button allows the new user to access the profile
>> details of the first user if the first user visited the page before
>> logging off.
>
> So, what you are saying is that the design of the web browser allows a
> second user to observe what the first user did by looking at the
> history and/or cache? There's not a lot you can do about that. You can
> send "no-cache" response headers to the browser, etc. but there's
> always a chance that the browser doesn't respect them, etc. and the
> history can be viewed.
>
> I'm not sure there's a way around that. Even if you use javascript to
> kill the window/tab, many browsers have a "re-open closed window/tab"
> that will resurrect the window/tab with the history in-tact, so you
> haven't bought anything there.
>
> I guess this is why you should be careful what you do from as public
> terminal, eh?
>
>> No secure data is held in the system.
>
> That's good, given the shaky security you've described here.
>
>> Can anyone suggest a way to change this? I am no expert on session
>> management.
>
> It's the browser that is the problem, not your session management. I
> think you need to instruct your users to completely exit the browser
> after they use your site if they value their privacy.
>
> - -chris
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.10 (MingW32)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
>
> iEYEARECAAYFAk54y0IACgkQ9CaO5/Lv0PAmLwCfRB69FXn1XUhPbMHQKD/Q/xAd
> QssAoJMKQk4xudqoGJlf0vkhdLZCkFkp
> =rYmn
> -----END PGP SIGNATURE-----
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
> For additional commands, e-mail: users-h...@tomcat.apache.org
>
>
> !DSPAM:4e78cb6c11371347337680!
>
>
>
--
Shanti Suresh
App Systems Analyst Lead
Web Services, LSA Development
University of Michigan
Office: 734-763-4807
sha...@umich.edu
http://lsa.umich.edu/cms
