Shibboleth Developers

[Paul Hethmon <>]

On 4/22/11 10:11 AM, "Christopher Bongaarts" 
<>
 wrote:

>>
>>In looking at my requirements, I'm thinking that hooking into the
>>storage service will be enough, but I'm looking for feedback in case I'm
>>missing something. My cluster requirements:
>>1. Hardware load balancer maintaining sticky sessions
>>2. Replication to other nodes accomplished within 5 minutes
>
>So you don't have any need for back-channel communication
>(attribute/artifact resolution, someday SLO)?

I don't need the back channel attribute/artifact resolution. I do need SLO
but I think my scheme will handle it, at least for front channel. That's
the reason for using the 2.1.5 release. Though I would like to strip out
the SLO support and package it separately so I can use newer Shib
versions. I do plan on making the replication delay time a configurable
option.

I suppose if my timing is not tight enough, I could add the filter to
trigger the replication action.

One of the things I've haven't spent enough time to figure out yet is how
the various Session objects fit together. I see on a typical login about 3
or 4 Session objects added to the storage service. One based on the
principal name, one on the session id, etc. Not sure yet whether those are
distinct objects are multiple references to the same object.

Paul