12-05-2022, 10:41 AM
I recently discovered some "problems" with deleting lots of objects in one go or importing a lot of objects in database mode
When a deletion process of user 1 is running all other users are blocked (by design of SQL Server locking behaviour) when starting ASGRD and loading initially or when pressing F5/refresh to update cache.
This runs into timeouts and error messages when the deletion process is taking longer than 30 seconds (default .NET timeout).
When setting the database to read commited snapshot endusers are able to retrieve data from before the deletion process started (which is often better to get potentially outdated results instead of an error message)
Just a thought, maybe there is something else that would imply not running the database in read committed snapshot, but this fixed my "problem" about users complaining that there app wouldn't start/refresh while I was working on the Connection objects.
I guess this is only coming up right now because I am doing a lot of work on the database while users are testing if the new setup is working as expected, but i thought it would at least be more of a suggestion as a feature request :-)
When a deletion process of user 1 is running all other users are blocked (by design of SQL Server locking behaviour) when starting ASGRD and loading initially or when pressing F5/refresh to update cache.
This runs into timeouts and error messages when the deletion process is taking longer than 30 seconds (default .NET timeout).
When setting the database to read commited snapshot endusers are able to retrieve data from before the deletion process started (which is often better to get potentially outdated results instead of an error message)
Just a thought, maybe there is something else that would imply not running the database in read committed snapshot, but this fixed my "problem" about users complaining that there app wouldn't start/refresh while I was working on the Connection objects.
I guess this is only coming up right now because I am doing a lot of work on the database while users are testing if the new setup is working as expected, but i thought it would at least be more of a suggestion as a feature request :-)
