Advanced Kubernetes cluster monitoring
With https://gitlab.com/gitlab-org/gitlab-ce/issues/38783 we could provide a good baseline and slightly beyond parity with the current k8s dashboard.
We have the opportunity to go much further, though:
- Track bandwidth used into and outside of the cluster
- Monitor individual deployments to detect pods that are requesting significantly more resources than they are actually using. This could be an area to optimize and get more efficiency out of your cluster. We can show a report here of the worst offenders by deployment, to simplify this.
- Monitor and display overall node and pod health. We could alert on flapping pods, pods that have repeatedly failed (and unlikely to succeed).
- Later on, we could also be interesting to show a resource usage chart, in short where are all my resources going? Perhaps we This could be a simple line or pie chart to display where the bulk of the resources are being consumed and by what deployment.
Edited by Joshua Lambert