Session Recovery after a shibd restart

[Scott Cantor]

> Anyways, I don't have any open questions, I just thought I would share what I learned (what I think I learned) and suggest that change.

That code is frozen, there aren't going to be any substantive changes and there are reasons that those kinds of changes wouldn't be viable anyway with that design.

The Apache copy is purely an optimization, it cannot allow use of sessions that don't exist in the real cache, and doing so would be pointless anyway since Apache is a multi-process server and you'd get inconsistent behavior.

As long as a longer term storage mechanism is used, restarting Apache or shibd has no effect on existing sessions (nor on RelayState, as it would be expected to be stored in the same place independent of both, if not using cookies to begin with.

The problem is that no longer term options are generally stable. They are, at least for some, stable in Java, and that is among many reasons for the redesign.

-- Scott