4-monitoring

Lab 4: Monitoring Functions

Kubeless monitoring relies on Prometheus. Each supported language runtimes are instrumented in such a way that they are automatically collecting metrics for each function invocation.

Grafana

You could also use Grafana to visualize the prometheus metrics exposed by Kubeless. Example of a Grafana dashboard for Kubeless showing function call rate, function failure rate and execution duration:

Expercise: Deploy a flakey function

Lets deploy a function that can generate some metrics. This function will fail on avegare 20% of the time – lets check if that holds true.

$ kubeless function deploy flakey-node \
  --runtime nodejs6 \
  --handler flakey.handler \
  --from-file flakey.js

Now lets generate some traffic for this function to populate the dashboard:

$ while :
> do
>   kubeless function call flakey-node --data 'test'
> done

Tutorial: View Metrcis

You can now go to Grafana and navigate to the Kubeless dashboard:

make open-grafana

Hint

Some time the built in dashboard does not load, try refreshing and if not create a new dashboard and upload the dashboard.json file inside this directory as the source.

Clean Up

When you are done you can exit your open shells and run the following command:

$ kubeless function delete flakey-node

---