Up until now we’ve been monitoring our managed RDS instances through Icinga, which has been working great. But for Kubernetes customers, we’ve been relying on Prometheus more and more to monitor external services, as well as the cluster itself. So it just makes sense that we also monitor RDS instances through it.
The way this monitoring works is very similar to others that we already have in place, like SQS or Elasticsearch monitoring. It uses the Prometheus cloudwatch-exporter to import metrics into Prometheus, for which we configure a set of alerts that will fire when something goes wrong.
This is already being rolled out to all those customers that run Kubernetes clusters and for which we manage their RDS instances. No action is required on your part.
For those customers that manage their RDS instances themselves, and still want to have this monitoring in place in their Kubernetes customers, get in touch with your lead engineer.