Starting with the release of Strategy ONE (March 2024), dossiers are also known as dashboards.

Description

The Strategy Metadata supports failover with proper configurations as outlined in this article . In the event of a primary Metadata Database Server failure during specific document manipulations, such as switching pages in Library consumption mode, the Library will continue to poll the I-Server for a period like10 minutes even after failover has occurred and the secondary Metadata DB server is operational.

Steps to reproduce

• Configure Intelligence Server against Metadata (eg., PostgreSQL). Make sure you can start/stop the PG Server.
  
• Enable the XML Trace log in Intelligence Server.
  
• Connect Strategy Library to the Intelligence Server, add a Dossier with multiple pages to your Library.
  
• In Library, execute the Dossier.
  
• Now stop the PostgreSQL Server of your Metadata. You can do this in the 'Services' panel of your Windows VM.
  
• Back to the Library, switch to another page of the Dossier. Notice there’ll be the following error in Library.
  
  

• Click OK, Library keeps sending polling command(from XML.log, you can see <grtex> commends being printed for 10 minutes), and further manipulations on this document will be blocked, like below

Root cause

When performing document manipulations, the Library follows a process where it first sends a <rbd> command to rebuild the document instance and then sends <grtex> commands to fetch the document instance JSON data. In the event that the database is down while the Intelligence Server is processing the <rbd> command, the <rbd> command will return the database error but will not fail the document instance. This situation can cause the Library to continuously poll the status, resulting in subsequent manipulations on this document instance hanging.

Workaround

During the hanging stage, user can go back to Library home and re-run the document. If the Metadata Database Server is back online at this time, everything will go well.
We’d also recommend you to apply the workarounds documented here to improve the Metadata failover performance: How to improve performance after Primary PostgreSQL Metadata Repository is down during Failover