I was thinking about Joe's question of how to invalidate the cache from the
user perspective... haven't fully looked into if/how this would be
implemented, but what about allowing methods in processors and controller
services to be annotated, indicating they should be exposed in the UI
somehow?

So there could be a clearCache() method in the distributed cache
service, annotated with something like @UserAction which then generates
an actions menu, or panel, in the UI where a user could click "Clear Cache".

In a lot of cases like ListHdfs, the processor stores state in a local file
and in the distributed cache (using the cache only when changing primary
nodes in a cluster). So the clear action would need to be done on the
processor in this case.

This is not to change the subject from Michael's suggestions of the
additional attributes and a MoveHdfs. Just wanted to throw the idea out
there.

On Thursday, December 3, 2015, Michael Dyer <[email protected]
<javascript:_e(%7B%7D,'cvml','[email protected]');>> wrote:

> Joe,
>
> Regarding the cache, I think that would need it's own processor.  Then I
> could specify the cache and which key (or all keys)...
>
> I would like to see GetHDFS / FetchHDFS with the same "Writes Attributes"
> as ListHDFS.  We're already picking up the file name and path, why not
> include the other 'hdfs.' namespaced values?
>
> What I would *really* like to have in this case is a simple 'MoveHDFS'
> processor that can operate off of the hdfs.values.
>
> Thanks,
>
> Michael
>
>
>

-- 
Sent from Gmail Mobile

Reply via email to