cas protocol behind an F5

[ROGERS Richard M]

On Fri, Sep 02, 2016 at 07:29:19AM +0000, ROGERS Richard M wrote:
> Not an F5, but we're currently going through a migration to Azure, and 
> started off with an Azure load balancer in front of our two IdP v3 
> servers. There were errors due to the whole logon session not being 
> performed on one server (traffic being shared). We've now changed the 
> load balancer to a traffic manager so that there's effectively one 
> live and one "hot standby" server, which seems to be working better.

Azuze load balancing doesn't support sticky sessions? I know F5 does, if that's what's causing the OP's problem. We have three active idp nodes behind a load balancer with no issues, but our load balancer is configured with sticky sessions so any given client is always routed to the same backend server for a given session (modulo that server becoming unavailable of course).

--
Paul B. Henson  |  (909) 979-6361  |  http://www.cpp.edu/~henson/ Operating Systems and Network Analyst  |  henson at cpp.edu California State Polytechnic University  |  Pomona CA 91768

-----

Yes, it does, but only per client (which means any back channel requests may not go to the same server, so potential for issues there which we decided to avoid - and we do have memcached running, but my understanding is that this may not be enough for some types of request).

Regards

Richard