non cluster IdP behind hardware load balancer

[Chris Phillips]

Hi Jean,

Keeping the configurations identical has been the practice I have used in
the past.  

Depending on change control practices and how you want to propagate
changes between instances you may choose to use SVN or Rsync as a way to
manage change and would allow you to recover from an 'oops' in the config.

Chris..




On 12-04-02 2:04 PM, "Jean Robertson" <jean.robertson at mcgill.ca> wrote:

>Hello,
>
>We are looking for a non-cluster pair of IdP's behind a hardware load
>balancer.
>
>The only reason we want it behind a load balancer is a fast recuperation
>in 
>case of failure.
>
>We are not looking to share load between servers.
>
>We are planning to have a single IdP server run for long periods of time
>and 
>should it fail, the standby would take over.
>
>We know that single sign on is lost in this case.
>
>We are using shibboleth IdP 2.1.5 software
>
>My question:
>- how to configure this?
>
>Do we call the AnonymousRelyingParty and DefaultRelyingParty the same in
>each 
>IdP's relying-party.xml config file? (The name being the one that the
>front 
>end is called)
>
>Do we copy the SHIBHOME/credentials cert and key from one machine to the
>other?
>
>Is there something I missed apart from these steps?
>
>I did look for documentation but did not quite find what is being
>proposed 
>here.
>
>Thanks,
> 
>Jean Robertson, McGill University
>
>--
>To unsubscribe from this list send an email to
>users-unsubscribe at shibboleth.net