tomcat-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Christopher Schultz <ch...@christopherschultz.net>
Subject Re: AW: mod_jk and session stickyness of images requests
Date Tue, 15 Dec 2009 18:12:08 GMT
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Timo,

On 12/15/2009 8:36 AM, Kockert, Timo wrote:
> Just to make it clear, here is a summary of my problem and what was suggested so far:

Thanks for the nice summary. After a long thread like this, it's nice to
have everything together.

> - In general the whole mod_jk load balancing works fine, with or
> without cookies (I tried disabling cookies on my iPhone and it worked
> as expected).

Good.

> - Some devices however don't seem to send cookies with image
> requests, although they send a valid cookie with the previous content
> request.

We'll be interested to see what those devices turn out to be. I suspect
that the iPhone is /not/ the culprit, since Safari is a pretty decent
web browser.

> - The latter may or may not be a problem of the devices themselves
> or an intermediate proxy or something else (we already had a problem with
> BlackBerry servers killing sessions).

Well, those servers can't kill the sessions, but they can interfere with
the cookies. One thing you could do, if you identify that certain user
agents ("browsers") are failing to handle the cookies properly, you can
disable cookies on certain browsers using a filter. If it comes to that,
I can help you write such a filter if such a thing doesn't already exist.

Can you describe the symptom in a bit more detail? Your claim is that
the request for the image goes to the wrong server in the farm. How are
you detecting that? Do you get an error when trying to access session
data that you expect to be there? Do you get mismatched session ids
(because the client requested session X, and the server didn't have that
session, so session Y was created on demand)?

I'm wondering if it's possible that there's a logic error in your code
that is actually masquerading as a backend server-switch and you might
be chasing the wrong problem.

Just a thought: I'm not saying your code sucks :)

- -chris
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAksn0XgACgkQ9CaO5/Lv0PBDxwCeLOTWWZ4SrB3H0edT3lHZM5nm
NMYAn34cfVpdCAgffca1823YQeeqmKBQ
=/ZQA
-----END PGP SIGNATURE-----

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org


Mime
View raw message