Can browser affect the session?

[HLady]

I posted earlier about this, I have an application and a sequence of
activities will cause the session to restart. My code will go to
session_start right after I execute a Response.Redirect(mypage) in
order to reload the page. I use this statement in other parts of the
code and it doesnt cause this. It is not a session.abandon statement,
or a timeout. Someone here suggested that the server recycling can
cause the session to restart, but I dont have anything in my event log
saying there was a recycle. This problem happens whether I use inproc
or stateserver session models.

What I found interesting, and this is new from my earlier post, is that
this problem with the unexplained session restart happens only on IE6.
I have tried the same thing on Opera and Firefox and it works
beautifully, no session restart happens. What I am wondering here is
how can the browser affect the session? Doesnt the session live in the
server, not understanding how this would make a difference.

Any ideas

 

[Marina Levit [MVP]]

The session does live on the server. However, the session cookie is on the
client - and the browser sends this up on every requests. I haven't
experienced what you are describing, but perhaps the session cookie is
getting lost somehow on the redirect.

 

[Robbe Morris [C# MVP]]

Some ISPs and corporate networks route their outgoing traffic
through proxy server farms. This can result in
session cookies getting reset because the
server thinks you are a new session from a
different ip.

You don't say much about the environment
you are experiencing this with. So, I'm
just throwing this out there for consideration.

 

[HLady]

Thanks for your suggestion, but this problem happens to my application
even when I am running the application locally on my machine as
localhost,

Another interesting thing is that the "session_end" method never runs,
I only get session_start method running again. This is when running
inproc session model, I know session_end doesnt run when using
stateserver, but should when running inproc. I get a different
sessionid too.