I'm excited about the promise of Open Mirroring so far it seems to be the best and cheapest (in capacity usage) solution to lift our on-prem SQL Server data to Fabric. However, I've come across a Major Issue that seems to be a bug.
When I setup the Mirrored Database in the Fabric Workspace all my testing works on the day the Database is created. The python script that reads the SQL Server CDC logs and writes the parquet files to the Mirrored Database Landing Zone works great and changes are reflected fairly quickly and smoothly.
BUT, when I attempt to execute this script the next day... the parquet files get written to the landing zone but the Mirrored Database does not process the changes into the Delta Tables and therefore the SQL Analytics Endpoing, also any additional attempts to make changes such as deleting files from the Mirrored Database through the Landing Zone do not get reflected in the database tables or the mirroring status pages in Fabric. Effectively the Database is frozen and does not respond to anything. The only thing I can do is delete it and create a new Mirrored Database.
The naming conventions and required files are all in the structure required from the documentation no additional files are being stored in the Landing Zone. I am adding another directory under the lakehouse Files location to store the CDC LSN's used for the last run, could this be an issue? But then why does it work on day 1 and not on day 2.
