Hi, I previously used query-frontend with a downstream_url pointing to my Prometheus instance, but that option was removed in a recent update.
Now I'm configuring query_frontend.address in Ruler to point directly to my Prometheus URL. However, alerts that use for: < duration > aren't working. For these alerts, the system sends an ALERTS_FOR_STATE query to Prometheus for the specific alert name - but that metric doesn’t exist in my Prometheus because all alerting rules in Ruler.
Everything worked as expected when I was using query-frontend with downstream_url. Is this behavior change intended, or is it a bug?