Database replication lag, you know, how do we keep an eye on it and what do we do when it gets bad?

were running a few read replicas for our main PostgreSQL instance and im noticing replication lag can sometimes creep up, especially during peak write times. we have some downstream services that depend on near real-time data, so a lot of lag is problematic. what strategies do you use to monitor this lag effectively? more importantly, how do you automatically react when it exceeds a certain threshold? are you buffering writes, failing reads, or something else entirely?