apache exporter stopped sending values for apache_exporter_scrape_failures_total on september 11th

We have an apache alert that's based on the metric apache_exporter_scrape_failures_total but from what I can see, there were two apache exporters exposing that metric and they both stopped exposing it at the same time on september 11th.

I couldn't find out exactly why they stopped. There's no change in puppet that seems to have the implication of changing the exporter's configuration. The apache alerts were added on the same day as when we stopped getting values for the metric.

Should we make the apache exporter expose apache_exporter_scrape_failures_total or should we rather remove the associated alert and unit test from prometheus-alerts?