#spinnaker pipelines

Kayenta is the Spinnaker microservice that is responsible for enabling canary deployments in Spinnaker and performing automated canary analysis. Canary deployment is a release deployment strategy where a new release version is deployed alongside the stable running version.

It is exposed to a small subset of the end-users and then evaluated against the baseline (a copy of current deployment) to find out if the new release is stable and ready to be rolled out. The application config in Spinnaker has the configuration option to activate canary config and canary reports on Deck.

Kayenta supports monitoring tools like Stackdriver, Prometheus, Datadog, New Relic, e.t.c. Performing automated canary analysis with Kayenta involves two stages — Metric Retrieval and Judgement. Key metrics are retrieved from canary and baseline clusters, tagged, and stored in a time-series database.

Slack is a popular communication and collaboration platform used widely in business for quick notifications and updates. Spinnaker supports sending CI/CD pipeline notifications to Slack.

In case a Spinnaker pipeline fails, DevOps teams could simply click on the link in a Slack notification and check for the reasons behind the failure. Spinnaker needs to be aware of the Slack bot’s identity (name & token) so that Spinnaker can communicate with the Slack bots and can automatically post notifications about Spinnaker pipeline information to a channel.

Canary Analysis is a deployment strategy for ensuring stable release where any change is partially rolled out and then evaluated against the current deployment i.e. baseline.

Kayenta microservice assesses the risk of a Canary release and checks for significant degradation between the baseline and Canary. Kayenta consists of two stages — Metric Retrieval and Judgement. Kayenta supports integration with APM monitoring tools like Stackdriver, Prometheus, Datadog, New Relic e.t.c. In the Metric Retrieval stage, Kayenta retrieves performance metrics from these tools. In the Judgement stage, it compares those retrieved metrics and makes a decision to pass or fail the canary release.