grpc

gRPC Go client with advanced settings

gRPC client, service and CLI utility to provide/check load balancing and connection stability

Using gRPC Go client

import "github.com/takama/grpc/client"
...

c, err := client.New(cfg, log)
if err != nil {
    log.Fatal(err)
}

ctx, cancel := context.WithTimeout(context.Background(), time.Second*15)
defer cancel()

e := echo.NewEchoClient(c.Connection())
response, err := e.Ping(
    // This is important to pass trough c.Context(ctx)
    // when we use Envoy balancer for any retries
    c.Context(ctx),
    &echo.Request{
        Content: "Some content",
    },
)
if err != nil {
    log.Print(err)
}

fmt.Print("Response:", response.Content)

Prepare service environment and configuration

• Create a new project and GKE cluster in Google Cloud
• Setup Contour balancer according to example
• Note somewhere the IP of Load balancer
• Change default GKE values in Makefile on your own

GKE_PROJECT_ID ?= your-project-id
GKE_PROJECT_ZONE ?= europe-west1-b
GKE_CLUSTER_NAME ?= your-cluster-name

• Choose domain host name for the service and point your DNS record on GKE Contour balancer using IP above
• Make the certificates for the domain name (Let's encrypt as option)
• Test to create in dry run grpc-service-tls in corresponded Kubernetes environment (ex: test)

kubectl create secret tls grpc-service-tls --key key.pem --cert cert.pem --dry-run -o yaml

• Create grpc-service-tls in corresponded Kubernetes environment (ex: test)

kubectl create secret tls grpc-service-tls --key key.pem --cert cert.pem

• Change domain name in .helm/values-test.yaml or any other values-name.yaml to your own

  ## Ingress route hosts
  ##
  hosts:
    ## gRPC service host
    - name: grpc
      host: grpc.your-test-domain.net
...
  ## Client connection to the service
  ##
  client:
    host: grpc.your-test-domain.net

• Change default namespace in Makefile

# Namespace: dev, prod, username ...
NAMESPACE ?= test

Build the service

make build

Deploy the service

make deploy

Build CLI utility on Linux

make compile

Build CLI utility on Mac OS

GOOS=darwin make compile

Run the utility with simple ping

./bin/darwin-amd64/grpc client ping --count 600 --config config/test.conf

Run the utility with reverse ping

./bin/darwin-amd64/grpc client reverse --count 600 --config config/test.conf

Scale services

During the ping process scale services from 1 to N and see how it is going

kubectl scale --replicas 3 -n test deploy/grpc
kubectl scale --replicas 6 -n test deploy/grpc
kubectl scale --replicas 1 -n test deploy/grpc
...

Other make commands

• all - run default complete set of commands (build the service)
• vendor - import all vendors (using dep)
• compile - build the service binary
• certs - download latests certs from an alpine image and prepare it for service container
• build - build container image
• push - push an image in docker registry
• run - build and run the service
• logs - show service logs from container
• deploy - deployment of the service into Kubernetes environment
• charts - validate helm templates (charts)
• test - run unit tests
• cover - show testing coverage for packages
• fmt - format Go packages with go fmt
• lint - use set of linters ( ~ 20) to check the service code
• stop - stop running container
• start - start existing container (if it was stopped before)
• rm - remove stopped container
• version - add next major/minor/patch version
• clean - remove binary and running container
• bootstrap - check and setup if something from utilities is not exist

Versioned changes

All changes in the project described in changelog

Generated using (Caldera boilerplate)