Kubernetes Opex Analytics 0.3.0

June 10, 2019. Kubernetes Opex Analytics 0.3.0 is released with the following enhancements:

• Enhancement #6 introducing Prometheus Exporter for consolidated metrics generated by Kubernetes Opex Analytics. So users can now configure a scrapper to integrate those metrics into their existing Prometheus monitoring environment. See the README to read more and learn how to getting started.
• Enhancement #5 introducing an integrated Grafana Dashboard along with the Prometheus exporter: this dashboard allows you easily get started with the Prometheus Exporter. This native dashboard aims to allow users to easily integrate Kubernetes Opex Analytics into their existing Prometheus/Grafana monitoring environment.

This two majors have been highly requested by users, so we hope that you will enjoy this release. Anyway don't hesitate to share your feedback!

The Team.

Kubernetes Opex Analytics v0.2.1

May 19, 2019. Pusblihing Kubernetes Opex Analytics 0.2.1, a patch version with the following changes:

• Fixed #12 related to the handling of default service account credential when RBAC is enabled: Kubernetes Opex Analytics pod now explicitly uses the default service account credential associated to the pod (See here to read more on default service account credential).
• Update frontend home title and charts' legends to highlight details related to selected cost model.
• Helm deployment: updated test pod configuration to wait a minute before running its check. This ensures that in case of image pulling the test will not run too early.

Update your deployment and enjoy.

Kubernetes Opex Analytics v0.2.0

May 12, 2019. We're pleased to announce that Kubernetes Opex Analytics version 0.2.0 is released. This version brings major enhancements, that have been made possible thanks to contributions from several people, that sent us feedback, submitted ideas for enhancements, and even made code and submitted pull requests (see the acknowledgement section for more details). All the changes introduced below are documented in README.md.

Introducing Cost Model

This is related to issue #7: by setting an estimated hourly cost for your Kubernetes cluster it's now possible to use actual costs instead of just cumulative usage for each period of time (daily and monthly). The hourly usage is still the ratio of resource usage per namespace.

To enable this capability, you have to set the following variables when launching Kubernetes Opex Analytics:

• KOA_COST_MODEL: Possible values are: CUMALITIVE_RATIO (default) indicates to compute cost allocation as cumulative resource usage of each period of time; CHARGE_BACK calculates cost based on given a cluster hourly rate; RATIO indicates to compute cost allocation as a normalized percentage of resource usage during the period of time.
• KOA_BILLING_HOURLY_RATE: sets a floating number corresponding to the estimated hourly cost of your Kubernetes cluster. For example if your cluster costs $5,000 a month (i.e. ~30 * 24 hours), its estimated hourly cost would be set to $6.95 = 5000/(30 * 24).
• KOA_BILLING_CURRENCY_SYMBOL (optional, default is $): sets the currency to annotate usage costs on charts.

Breaking Changes: This feature required to update the internal data model of Kubernetes Opex Analytics, what leads to breaking compatibility with version 0.1.0. Unfortunately we won't be able to provide effort on a migration tool, considering pending features we have in backlog. Hence you should thus either keep your current version, or start the new version from a clean data path (by setting the configuration variable KOA_DB_LOCATION accordingly).

To conclude on this feature, we want to give claps to @luisdavim that contributed with interesting ideas.

New Configuration Variables

• KOA_K8S_API_VERIFY_SSL (boolean, default: false): enables to pass insecure flag when calling Kubernetes API (e.g. when using self-signed certificate). It should be set to true to enable insecure SSL connection. Thanks to @luisdavim to have contributed this feature (PR #9).
• KOA_ENABLE_DEBUG: when set to true it enables the data collection and analytics backend to run in debug mode, enabling more logs as well as additional features for debugging.
• KOA_K8S_AUTH_TOKEN: taken into account in debug mode, it enables to set an authentication token for Kubernetes API. See here to learn how to get an authentication token for Kubernetes API. Warning: Note that in production it's not recommended to bind security token through environment variables.

Helm Deployment

It's now possible to deploy Kubernetes Opex Analytics using Helm (see the README.md for more details).

Thanks to @luisdavim (yes, the same) that contributed this feature (PR #9).

Chart Rendering

• Improved analytics data export strategy to cut down the time needed to have all charts available for the first time.
• Updated number format on node usage legends, switching from the percent notation to raw values.

Acknowledgements

First, we want to thank people that mind to give us a star to welcome our project, it's really motivating.

Then, we want to say a special thanks to, @pousa, @nelvadas, and @luisdavim, who taken care to share their feedback, to open issues, and/or to contribute pull requests.

v0.1.0

First release