Add DORA metrics

[fsequeira1]

The idea behind this is to add support for some DORA metrics:

Deployment Frequency — How often an organization successfully releases to production

flagger_count_canary_success - flagger_count_canary_success offset 24h / 24  # deployments in a day

Change Failure Rate — The percentage of deployments causing a failure in production

100 * ( flagger_count_canary_failure / flagger_count_canary_success + flagger_count_canary_failure )

I'm wondering if there is any way to grab the amount of time it takes a commit to get into production and how long it takes to recover from a failure in production.
However, I'm unsure if I can do it with the current metrics.

[stefanprodan]

I'm wondering if there is any way to grab the amount of time it takes a commit to get into production

The histogram could be used to extract this, also the two metrics that you've added, I think they can be composed from the existing histogram count metric.

[fsequeira1]

I assumed the same, however, I couldn't get the results I wanted, maybe I'm doing something wrong. I want to know the number of times there is a failure, which is different from the number of failures that you can see in the following picture:

Can you elaborate/point me to how I could use the histogram to extract that info? I didn't follow

[stefanprodan]

The histogram I'm referring to is called canary_duration_seconds, please try that.

[fsequeira1]

I think this will solve the issue with one of the metrics (Lead Time for Changes). However the other seems to be dependent on external factors (the deployment can be healthy and the application failing) and probably the best way is to collect it from some kind of external checker like blackbox-exporter.

Lead Time for Changes (from a deployment point of view) — The amount of time it takes a commit to get into production

rate(gotk_reconcile_duration_seconds_sum[5m])/ rate(gotk_reconcile_duration_seconds_count[5m]) > 0
+
rate(flagger_canary_duration_seconds_sum[5m]) / rate(flagger_canary_duration_seconds_count[5m]) > 0