Porting advanced v2 resolver config to v3

[Cantor, Scott]

On 2/11/15, 2:42 PM, "Marvin Addison" <marvin.addison at gmail.com> wrote:


>
>I updated to last night's snapshot to pull in additional logging and 
>noticed the following entry 
>
>2015-02-11 09:31:49,946 - TRACE 
>[net.shibboleth.idp.attribute.resolver.dc.impl.AbstractSearchDataConnector
>:179] - Data Connector 'edEntitlement': Cache found, resolved attributes 
>null using cache 
>com.google.common.cache.LocalCache$LocalManualCache at 579736d4

I fixed some major bugs in the results caching, but AFAIK it should be 
working now in the snapshot.

>I have a cache bean defined in the Spring XML file sourced for _each_ 
>data connector, the dependency and the dependent, and these are separate 
>bean instances judging from the following entry, which appears earlier in 
>the log for the same request:

That's certainly been the norm, since the old way involved using the 
ResultsCache element, not a bean. But if you had the same XML file defined 
to each connector, what I would expect is that each one produces its own 
child context (given Rod's fix at least).

There's no way to share the beans that way. To do that, you'd need to put 
the bean into the root context via global.xml I think, or you could load 
it in an additional resource loaded into the resolver service.

>Shouldn't I be using a shared cache? I have no idea how to do that, 
>however, unless there's a shibboleth.Cache bean in the root context that 
>I missed.

We added a ResultCacheBean element (or whatever it's called) to support 
that option.

-- Scott