hi, I am trying to understand the failure and recovery modes in airbyte. • I have a source with 100 tables, and need to sync them all to one distination. • The sync can take many hours. • From time to time a single table will fail e.g. due to schema issues etc. • Desired behavior: ◦ Single connection for the full 100 tables ◦ Job attempt to sync all tables ◦ Failure in one table doesn’t stop the job - but the result of the job will contain the tables that failed ◦ Ability to “retry” only the failed tables - including manually Is that possible? Or is it that a single connection succeeds, fails and retries monolitically?

It depends on the type of failure but yeah, the current behaviour is that generally if a stream does not sync then the entire sync will fail. To some extent this is by design, to ensure that all expected data is delivered "at least once" into the destination, but there may be room for improvement - could you submit a feature request with your suggestion?