We have an app database that replicates to a reporting database via a complex trigger-based ETL process. For various reasons, some records are missing important data. Currently the ETL process is the only connection with write access. We want to insert the missing data directly rather than attempt to rebuild the triggers, and we can make that an atomic operation. However, it is possible that the ETL may be updating the same record. Is there a risk of interrupting the replication process and/or causing performance issues? Everything is mysql.