Pg_exporter_last_scrape_error provides 1 on crunchy cluster replica

Hi. I’m using the postgres_exporter for my crunchy cluster. I have 3 instances of one database running. 1 leader and 2 replicas. I have configured an alert pg_exporter_last_scrape_error > 0 to fire if pg_exporter_last_scrape_error is greater 0 for 60s. I get an alert when the alert query is executed on a replica. Is this an error of the postgres_exporter or should the expression only be performed on the leader of the cluster? How can the expression be executed only on the leader instance?
Thanks for your help!