IDP node stickiness without an SLB?
David Langenberg
davel at uchicago.edu
Tue Jan 21 21:22:30 EST 2014
Please let us know what worked. I'll be attempting something similar soon &
it'd be nice to know what worked for you.
Dave
On Tuesday, January 21, 2014, Wessel, Keith <kwessel at illinois.edu> wrote:
> All,
>
> Thanks, some good suggestions.
>
> Chris, I wish your idea would work as it's so easy to implement. Scott's
> correct, though, and we saw exactly that behavior with our GSLB. A user
> starts on node 1 and, once being redirected to the UsernamePassword login
> handler, switched to the other node... and things break there.
>
> We'll experiment with both Dave's keep-alive idea and, if not, look at the
> host cookie and proxy idea.
>
> Thanks again, everyone.
>
> Keith
>
>
> -----Original Message-----
> From: users-bounces at shibboleth.net <javascript:;> [mailto:
> users-bounces at shibboleth.net <javascript:;>] On Behalf Of Cantor, Scott
> Sent: Tuesday, January 21, 2014 7:50 PM
> To: Shib Users
> Subject: Re: IDP node stickiness without an SLB?
>
> On 1/21/14, 6:38 PM, "Christopher Bongaarts" <cab at umn.edu <javascript:;>>
> wrote:
> >
> >Off the top of my head - not sure if this would actually work:
> >
> >Update the login.jsp to look at the Host: header - if it's the "virtual"
> >name (idp.illinois.edu), redirect to one of the "real" names
> >(idpN.illinois.edu). Otherwise leave it alone.
> >
> >I'm pretty sure that you only really need to sticky that page (because
> >the login context stored locally to that IdP server).
>
> I don't think so. By the time login.jsp is involved, the login context has
> been created and a cookie sent back, and a switch there would break it.
> Same goes for after the password is posted, I think, but I'm less sure of
> that step, that may be a part of my customizations.
>
> At least in my IdP, you have /.../SAML2/SSO/Redirect -> /idp/AuthnEngine
> -> /.../SAML2/SSO/Redirect -> SP, and breaking the sticky at any stage of
> that will break the login.
>
> I think you'd have to do something like what Jim described.
>
> -- Scott
>
>
> --
> To unsubscribe from this list send an email to
> users-unsubscribe at shibboleth.net <javascript:;>
> --
> To unsubscribe from this list send an email to
> users-unsubscribe at shibboleth.net <javascript:;>
>
--
David Langenberg
Identity & Access Management
The University of Chicago
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://shibboleth.net/pipermail/users/attachments/20140121/40684a8c/attachment.html
More information about the users
mailing list