Closed samber closed 4 years ago
Note that this is not an exporter, it's just a metric library. Users are expected to configure metrics according to their needs.
For a few basic metrics mentioned in the README (request count and latency) you would probably want to configure alerts based on error ratio and percentile latency. For example:
- alert: NginxTooMany500s
expr: sum by(host, node) (rate(nginx_http_requests_total{status=~"5.."}[30m]))
/ sum by(host, node) (rate(nginx_http_requests_total[30m])) > 0.01
annotations:
text: Nginx for {{ $labels.host }} on {{ $labels.node }} serving more than 1% errors
- alert: NginxLatencyHigh
expr: histogram_quantile(0.99,
sum(rate(nginx_http_request_duration_seconds_bucket[30m])) by (host, node, le))
> 10
annotations:
text: Nginx p99 latency for {{ $labels.host }} on {{ $labels.node }} is higher than 10 seconds
I hope this helps.
Hi there 👋
I am looking for alerting rules that fit with this exporter, in order to feed an "awesome" collection: samber/awesome-prometheus-alerts.
The goal is to offer "out-of-the-box" rules in a single place.
Anybody here has such a thing ? 🙏
I think we should cross-commit in this repo as well!