Open samsja opened 4 years ago
Greetings,
I've encountered same issue with controlling stackdriver_exporter
rate.
Turns out that --monitoring.metrics-interval=5m
is used to collect metric data points for that interval (in particular example all data points for last 5m). Name of the arg is somewhat confusing.
Another find is that stackdriver_exporter
firing GCP API calls each time it queried otherwise exporter stays idle. In other words: each time prometheus
scraping stackdriver_exporter
the exporter at same moment firing GCP API calls. So the rate is should controlled by prometheus scrape_interval
option. In your case scrape_interval: 1h
.
Hope it helps.
I installed the stackdriver_exporter (v0.6.0) with helm on my K8s cluster. I only want to monitor my cloud sql DataBase. I set 1h interval (stackdriver.metrics.interval = 1h) in the values.yaml.
However when I call the stackdriver_monitoring_api_calls_total expressions with prometheus, I see a rate of 4500 calls pers hours.
More anormal when I look closely, i see that each 30 seconds, there are 40 more api calles.
Is that the normate rate of stackdriver_exporter ? Shouldn't I see new api call only each hours ?
thanks in advance