On Feb 7, 2008, at 12:20 PM, Robert Leland wrote:

Scott Ferguson wrote:


On Feb 7, 2008, at 8:51 AM, Robert Leland wrote:


In a single machine clustered environment the 20 minute timeout works, but in a two machine cluster the session does not timeout until 30 minutes..

Actually, that should be normal.

To reduce network traffic, the active times are only updated infrequently, and therefore the timeout is treated as a minimum that gets stretched. So you won't see the sessions closed at exactly the session timeout value, but some minutes later.


We have had a live session after 25 minutes....
So in order to have a 95% chance that the session times out at 20 minutes we have to set a session-timeout to .... what 15 minutes ?
That seems a little counter intuitive to the spec definition.

Well, the spec is never going to specify that a session has a timeout of exactly 20 minutes, since that's impossible to implement. It does guarantee that the session will live for at least 20 minutes.

In theory, we could add a session-update-window as a configuration parameter. I'm not sure I understand why that would be important to an application, though.

-- Scott



I would have liked the answer better if it was "its an issue that has been improved/fixed in the latest version of resin" then I could have
encouraged my client to buy new licenses :) !



-- Scott



The only difference in the single vs. two machine resin.conf is:

Single machine
//Use caucho directly no ISAPI filer
<http server-id="app1" host="127.0.0.1" port="80" read- timeout="300s" write-timeout="300s"/>
//One server in cluster
<srun server-id="app1" host="127.0.0.1" port="6802" index="1"/>
        <web-app-deploy path="webapps"/>
//Runs in development directory
<web-app id="/xxxx" document-directory="D:/Projects/IOP/XXXX/ code/web">
      </web-app>
<web-app id="/xxxx-help" document-directory="webapps/xxxx- help">
      </web-app>


Two machine Cluster (No front end load balancer)
//Uses IIS via ISAPI as front end no direct access to resin configured.
        <srun server-id="app1" host="web1" port="6802" index="1"/>
        <srun server-id="app2" host="web2" port="6802" index="2"/>
    //Uses directory  with war files
     <web-app-deploy path="d:/web/apps/war-dir"/>


I couldn't tell if this issue might have been fixed my 3.1.4 or 3.0.25,
we are currently using 3.0.23.
Possibly related Issues that were fixed.

security: update form-login with session timeout issue (#1674, rep by bthakkar) session saving was still attemped after invalidation (#2031, rep by jkriesten) timing issues with session invalidation (#1897, rep by Bradley Van Cleave) session: update invalidation to ensure invalidation drops session (#1951)





 -Rob
_________________________________________
Robert Leland                  INTEGRITY One   Partners
P: (703) 581-6522 1900 Campus Commons Drive, Suite 150
F: (703) 476-7405                          Reston, VA 20191
[EMAIL PROTECTED]

BUSINESS CONSULTING | TECHNOLOGY | INNOVATION R&D

_______________________________________________
resin-interest mailing list
resin-interest@caucho.com
http://maillist.caucho.com/mailman/listinfo/resin-interest


_______________________________________________
resin-interest mailing list
resin-interest@caucho.com
http://maillist.caucho.com/mailman/listinfo/resin-interest



--
 -Rob
_________________________________________
Robert Leland                  INTEGRITY One   Partners
P: (703) 581-6522 1900 Campus Commons Drive, Suite 150
F: (703) 476-7405                          Reston, VA 20191
[EMAIL PROTECTED]

BUSINESS CONSULTING | TECHNOLOGY | INNOVATION R&D

_______________________________________________
resin-interest mailing list
resin-interest@caucho.com
http://maillist.caucho.com/mailman/listinfo/resin-interest

_______________________________________________
resin-interest mailing list
resin-interest@caucho.com
http://maillist.caucho.com/mailman/listinfo/resin-interest

Reply via email to