Retrieval

The NewsGator News Rapid Retrieval Job retrieves all new content from credentialed feeds that have frequently updating content, such as Twitter, Yammer, SalesForce Chatter, and Dynamics CRM.

The NewsGator News Retrieval Job is doing the aggregation of all other feed content. The service attempts to update every feed in the system within one hour. Each time this job runs, it attempts to retrieve all new content from a fraction of the total feeds in the system.

For example, if this job runs every 15 minutes, it attempts to retrieve one quarter of all the feeds each time it runs.

Retrieving a great number of feeds creates significant load. While sizing is dependent on hardware (and to some degree, the nature of the feeds being retrieved), we recommend that you do your initial deployment with 500 or fewer feeds, a fifteen-minute retrieval cycle, and two threads.

The second box in the Job Configuration section lets you configure the number of threads allowed for news retrieval. This setting affects the tradeoff between time to retrieve and the overall load placed on the server.

Retrieval behavior: older posts on live feeds are not retrieved

The following is not a configurable behavior. It is described here to provide understanding.

In some situations, an article can be created with a published date that is in the past. Prior to the 3.1 release, News Stream would not create activity stream items from such articles to avoid publishing out-of-date news into the stream.

Since the 3.1 release, if the publish date is within 72 hours of the current time, News Stream adds the article to the stream.

For SharePoint feeds that do not have a publish date within 72 hours of the current time, News Stream looks at the description of the article for fields that may contain the publish date. If there is a publish date within 72 hours in a publish date field in the description, News Stream adds the article to the stream.

The publish date of the article is kept as it appeared in the source feed (for example, if an article has a published date that is 30 hours in the past, it is added to the stream and the published date is kept at 30 hours in the past).