Captive portal logouts



  • Hi,

    Have just reinstalled a server from an early 2.0 Alpha-alpha version to 2.0-RC3 (June 21st build) that is providing a captive portal.  Everything was working perfectly and was a huge improvement from the alpha-alpha version where the portal login page would fail to appear regularly etc.

    I've just done an auto-update to July 4th snapshot and now I'm finding my portal clients are being logged out after a few seconds.  The idle timeout is set to 120 mins, with hard timeout set to 420.

    Restarting the captive portal service makes no difference - here's the portal authentication logs:

    Jul 8 09:57:01 	logportalauth[21921]: LOGIN: testuser, , 10.5.192.14
    Jul 8 09:57:32 	logportalauth[14166]: TIMEOUT: testuser, , 10.5.192.14
    Jul 8 10:01:42 	logportalauth[21921]: LOGIN: testuser, , 10.5.192.14
    Jul 8 10:02:34 	logportalauth[10568]: TIMEOUT: testuser, , 10.5.192.14
    Jul 8 10:09:56 	logportalauth[19176]: LOGIN: testuser, , 10.5.192.14
    Jul 8 10:10:35 	logportalauth[41891]: TIMEOUT: testuser, , 10.5.192.14
    Jul 8 10:24:07 	logportalauth[63680]: Restarting captive portal.
    Jul 8 10:24:18 	logportalauth[63358]: Restarting captive portal.
    Jul 8 10:26:00 	logportalauth[20421]: LOGIN: testuser, , 10.5.192.14
    Jul 8 10:26:19 	logportalauth[15393]: TIMEOUT: testuser, , 10.5.192.14
    

    Anyone got any ideas?

    Thanks,
    Stuart



  • Upgrade to latest snapshot the issue has been fixed.



  • @ermal:

    Upgrade to latest snapshot the issue has been fixed.

    Thanks - I actually saw the sticky thread about using the RELENG snapshots after I had posted.  At the time, I was under the impression that I was on the latest version at the time, but am now running the snapshot from yesterday.

    Fingers crossed, everything is looking good at the moment.

    Regards,
    Stuart



  • Stupots, if you want to see the bug report for this issue, take a look at : http://redmine.pfsense.org/issues/1653

    A patch that adds the functionality to report an ip address's last activity timestamp was left out of the build.  So idle timeouts always took effect immediately.
    Josh


Locked