Skip to content

Latest commit

 

History

History
84 lines (58 loc) · 3.05 KB

README.md

File metadata and controls

84 lines (58 loc) · 3.05 KB

K8s manager

Description

This service is part of the Engine, exposes a gRPC service to encapsulate all Kubernetes related features and Prometheus queries to get metrics and alerts. The only service that is going to call this gRPC is the Admin API service when need to create new Kubernetes resources.

gRPC

The Protobuf file and the code generated are within proto folder.

To generate the code from the .proto file run the following command.

./scripts/generate_proto.sh

We expose the following services in the gRPC server:

  • RuntimeService: This service manage the runtime lifecycle and expose two functions:

    • Create: to create a new runtime
    • RuntimeStatus: to get the status of a runtime, in order to return to the front when is ready.

  • VersionService: Is intended to control the versions lifecycle with the following functions.

    • Start
    • Stop
    • Publish
    • Unpublish
    • UpdateConfig

  • ResourceMetricsService: Once a runtime and a version are running with this service we return to the Admin API the resource metrics stored in Prometheus. Expose the following functions.

    • GetVersion: return an array of metrics of CPU and Memory with a timestamp.
    • WatchVersion: create a gRPC stream to return realtime metrics to the Admin API in order to update graphs in the front.

Kubernetes

This server uses the official kubernetes sdk client-go to interact with the cluster.

We create these resources:

  • Namespace: Each Runtime is deployed in its own namespace with a matching name.

  • RBAC: The Kubernetes objects Role, ServiceAccount and RoleBinding to allow the K8s Runtime Operator to create the required objects within the namespace.

  • K8s Runtime Operator: This is the Kubernetes operator that extend the Custom Resource Definition to create the Runtime resource.

  • Runtime: This custom resource is the base infrastructure needed to later run a RuntimeVersion created with a .krt file.

Prometheus

Create a manager to handle all the required queries to Prometheus to print the graphs in the front.

Before launch the service you need to open a port-forward to Prometheus service and update the config.yaml with the Prometheus URL pointing to localhost:9090. To open the port-forward run the below command.

kubectl -n kre port-forward svc/prometheus-prometheus-oper-prometheus 9090:9090

Once the service have started this service in your local environment there are a script to perform gRPC calls that trigger the Prometheus queries (./scripts/metrics-grpc-call.sh). The script output should show something like below.

$ ./scripts/metrics-grpc-call.sh
{
  "versionResourceMetrics": [
    {
      "date": "2020-06-11 15:56:48 +0200 CEST",
      "cpu": 0.00024974045489303945,
      "mem": 56677.52937041498
    },
    {
      "date": "2020-06-11 16:01:48 +0200 CEST",
      "cpu": 0.0003369122322523001
    },
[...]