Dear @samantbl,
Regarding @tobias.koetter’s suggestion to re-execute the DB nodes on failure, here you can find an example how to do so: