[POC]Automating the deployment of kube-state-metrics during K8s Observability onboarding
gizas opened this issue · comments
Andrew Gizas commented
Goal
As part of the Observability Onboarding initiative for kubernetes, kustomise has been referenced as a possible solution to ease the onboarding of our users through Kibana UI. Basic goals for the use of kustomise templates are:
- Support the already existing installation of managed/standalone manifests with Kubernetes integration being installed and ready to collect logs/metrics from a Kubernetes cluster
- Support the installation of kube-state-metrics library that is needed for the metrics collection
Deliverables
As part of this POC will try to:
- Provide the needed kustomise templates to support both managed and standalone elastic agent scenarios
- Investigate if it is possible users to call online versions of kustomise templates in order to eliminate the need to download any files in their environment before running any commands
- Replace the secrets that are inside elastic-agent manifests with reference to actual kubernetes secrets that will be preconfgured inside the kubernetes cluster. As part of the onboarding we will provide the commands needed to create those secrtes in Kubernetes environment
- Provide (when/if possible) one-liner commands (commands that will be copy-pasted in the shell directly)
- [Stretched goal]: Provide a processor in elastic-agent standalone templates that will enhance the ingested documents with the new
token-id
. This will allow us later in kibana side to check if agents with specificuser-provided id
have been able to ingest data successfully
References
- #2175
- #2213
- https://kubernetes.io/docs/tasks/configmap-secret/managing-secret-using-kustomize/
- Current kustomise templates for elastic-agent (managed and standalone)
- elastic/kibana#181213 - The CI automation that is based in Makefile. Integrates the ts files needed for Kibana to be able to create the Kubernetes manifests during onboarding
Miguel Luna commented
Let's make the title aligned with the goal of the POC which is automating the deployment of kube-state-metrics during onboarding