The previous logic deleted payload data that was associated with the rows in the History table. In theory, this should "work" because there shouldn't be any payload data (besides custom status) that doesn't have an associated record in the History table. However, it seems that this isn't necessarily the case, or that we're losing track of certain payload IDs over time.
Regardless of the root cause for the previous issue, it makes sense to just purge all runtime data as part of ContinueAsNew. This PR does exactly that, ensuring that we can't run into this kind of problem in the future.
Resolves https://github.com/microsoft/durabletask-mssql/issues/100
The previous logic deleted payload data that was associated with the rows in the History table. In theory, this should "work" because there shouldn't be any payload data (besides custom status) that doesn't have an associated record in the History table. However, it seems that this isn't necessarily the case, or that we're losing track of certain payload IDs over time.
Regardless of the root cause for the previous issue, it makes sense to just purge all runtime data as part of ContinueAsNew. This PR does exactly that, ensuring that we can't run into this kind of problem in the future.