Enterprise Delivery Basics

10 examples to get you started with Enterprise Delivery - 7 basic and 3 intermediate.

Prerequisites

Install Go 1.26.x or later from go.dev/dl.
Create a module: mkdir deliverylab && cd deliverylab && go mod init example.com/deliverylab.
Optional for flag and metrics examples: go get github.com/open-feature/go-sdk/openfeature github.com/prometheus/client_golang/prometheus.

Basic Examples

1. Embed Build Version at Link Time

Stamp git SHA into the binary for deploy correlation and /healthz responses.

package main
 
import (
    "fmt"
    "runtime/debug"
)
 
func main() {
    fmt.Println("version:", buildVersion())
}
 
func buildVersion() string {
    info, ok := debug.ReadBuildInfo()
    if !ok {
        return "dev"
    }
    for _, s := range info.Settings {
        if s.Key == "vcs.revision" {
            if len(s.Value) >= 7 {
                return s.Value[:7]
            }
            return s.Value
        }
    }
    return "unknown"
}

debug.ReadBuildInfo works when built with module info (go build from a git repo).
Pass -ldflags "-X main.version=..." for explicit stamping in CI.
Log and expose this value on every deploy for rollback decisions.

Related: Shipping Go Services Safely at Enterprise Scale - delivery overview

2. CI Test Gate Script

A minimal pipeline step that fails the build on test or race failures.

#!/usr/bin/env bash
set -euo pipefail
go version
go test ./... -count=1 -race -timeout 5m
go vet ./...

-race catches data races before prod - worth the CI time on service code.
-count=1 disables test cache so CI always re-runs.
Wrap this in GitHub Actions, GitLab CI, or Buildkite before image build.

Related: DORA Metrics for Go Teams - measuring pipeline health

3. Multi-Stage Dockerfile for Reproducible Images

Compile in a builder stage; ship a minimal runtime image.

FROM golang:1.26-bookworm AS builder
WORKDIR /src
COPY go.mod go.sum ./
RUN go mod download
COPY . .
RUN CGO_ENABLED=0 GOOS=linux go build -trimpath -ldflags="-s -w" -o /out/app ./cmd/api
 
FROM gcr.io/distroless/static-debian12
COPY --from=builder /out/app /app
USER nonroot:nonroot
ENTRYPOINT ["/app"]

CGO_ENABLED=0 yields a static binary suitable for distroless.
-trimpath removes local paths from binaries for reproducibility.
Tag the built image with git SHA, not :latest, in your registry.

Related: Release Strategy: Blue-Green & Canary for Go - deploying images

4. Readiness Probe Handler

Separate cheap liveness from dependency-checked readiness.

package main
 
import (
    "database/sql"
    "net/http"
    "os"
 
    _ "github.com/jackc/pgx/v5/stdlib"
)
 
func main() {
    db, _ := sql.Open("pgx", os.Getenv("DATABASE_URL"))
    mux := http.NewServeMux()
    mux.HandleFunc("GET /healthz", func(w http.ResponseWriter, _ *http.Request) {
        w.WriteHeader(http.StatusOK)
    })
    mux.HandleFunc("GET /readyz", func(w http.ResponseWriter, _ *http.Request) {
        if err := db.Ping(); err != nil {
            http.Error(w, "db down", http.StatusServiceUnavailable)
            return
        }
        w.WriteHeader(http.StatusOK)
    })
    http.ListenAndServe(":8080", mux)
}

Liveness should not call external systems - only readiness should.
Return 503 on readiness failure so load balancers stop traffic during rollout.
Kubernetes uses these paths in livenessProbe and readinessProbe.

Related: Progressive Delivery & Automated Verification - automated checks

5. Graceful Shutdown for Rolling Deploys

Drain in-flight requests before SIGKILL during pod termination.

package main
 
import (
    "context"
    "log/slog"
    "net/http"
    "os"
    "os/signal"
    "syscall"
    "time"
)
 
func main() {
    srv := &http.Server{Addr: ":8080", Handler: http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) {
        time.Sleep(2 * time.Second)
        w.WriteHeader(http.StatusOK)
    })}
    go func() { _ = srv.ListenAndServe() }()
 
    quit := make(chan os.Signal, 1)
    signal.Notify(quit, syscall.SIGTERM, syscall.SIGINT)
    <-quit
 
    ctx, cancel := context.WithTimeout(context.Background(), 15*time.Second)
    defer cancel()
    if err := srv.Shutdown(ctx); err != nil {
        slog.Error("shutdown", "err", err)
    }
}

Kubernetes sends SIGTERM before removing endpoints from Service.
Shutdown waits for active handlers up to the timeout.
Match terminationGracePeriodSeconds to your worst-case request duration.

Related: Rollback Runbooks for Go Deployments - recovery during bad rollouts

6. Environment-Based Config (12-Factor)

Same binary across staging and prod; config from environment only.

package main
 
import (
    "log/slog"
    "os"
)
 
type Config struct {
    Port        string
    DatabaseURL string
    LogLevel    string
}
 
func loadConfig() Config {
    return Config{
        Port:        envOr("PORT", "8080"),
        DatabaseURL: os.Getenv("DATABASE_URL"),
        LogLevel:    envOr("LOG_LEVEL", "info"),
    }
}
 
func envOr(key, def string) string {
    if v := os.Getenv(key); v != "" {
        return v
    }
    return def
}
 
func main() {
    cfg := loadConfig()
    slog.Info("starting", "port", cfg.Port, "log_level", cfg.LogLevel)
}

Never bake environment-specific values into the image at build time.
Validate required vars at startup and fail fast with clear logs.
Secrets belong in a secret manager, referenced as env vars in the pod spec.

Related: Change Management & Compliance Gates - controlled prod changes

7. Post-Deploy Smoke Test

A tiny Go test binary CI runs against staging after deploy.

package main
 
import (
    "fmt"
    "net/http"
    "os"
    "time"
)
 
func main() {
    base := os.Getenv("SMOKE_BASE_URL")
    if base == "" {
        base = "http://localhost:8080"
    }
    client := &http.Client{Timeout: 5 * time.Second}
    for _, path := range []string{"/healthz", "/readyz"} {
        resp, err := client.Get(base + path)
        if err != nil || resp.StatusCode != http.StatusOK {
            fmt.Fprintf(os.Stderr, "smoke failed %s: %v status=%v\n", path, err, resp)
            os.Exit(1)
        }
        resp.Body.Close()
    }
    fmt.Println("smoke ok")
}

Run smoke immediately after promote to staging; block prod if it fails.
Extend with authenticated happy-path API calls for critical flows.
Keep smoke fast (under 30 seconds) so pipelines stay snappy.

Related: Progressive Delivery & Automated Verification - verification depth

Intermediate Examples

8. Feature Flag Interface (Dark Launch)

Deploy new code with behavior off until ops enables the flag.

package main
 
import (
    "context"
    "os"
)
 
type Flags interface {
    Bool(ctx context.Context, key string, defaultVal bool) bool
}
 
type EnvFlags struct{}
 
func (EnvFlags) Bool(_ context.Context, key string, defaultVal bool) bool {
    if v := os.Getenv("FLAG_" + key); v == "true" {
        return true
    }
    if v == "false" {
        return false
    }
    return defaultVal
}
 
func runNewCheckout(ctx context.Context) error    { return nil }
func runLegacyCheckout(ctx context.Context) error { return nil }
 
func handleCheckout(ctx context.Context, flags Flags) error {
    if flags.Bool(ctx, "NEW_CHECKOUT", false) {
        return runNewCheckout(ctx)
    }
    return runLegacyCheckout(ctx)
}

Start with env-backed flags for simplicity; graduate to OpenFeature or LaunchDarkly.
Default new flags to false in production config.
Dark launch lets canary pods run new paths without user exposure.

Related: Feature Flags in Go Services - production flag SDKs

9. Prometheus Deploy Info Metric

Expose version labels so dashboards mark deploy boundaries.

package main
 
import (
    "net/http"
    "runtime"
 
    "github.com/prometheus/client_golang/prometheus"
    "github.com/prometheus/client_golang/prometheus/promhttp"
)
 
var buildInfo = prometheus.NewGaugeVec(
    prometheus.GaugeOpts{Name: "go_build_info", Help: "Build metadata"},
    []string{"version", "goversion"},
)
 
func main() {
    prometheus.MustRegister(buildInfo)
    buildInfo.WithLabelValues("abc1234", runtime.Version()).Set(1)
    http.Handle("/metrics", promhttp.Handler())
    http.ListenAndServe(":9090", nil)
}

Register metrics once in main, never per request.
Grafana annotations can watch go_build_info label changes for deploy lines.
Pair with RED charts to spot CFR spikes after version label changes.

Related: DORA Metrics for Go Teams - change failure rate

10. Expand-Contract Migration Guard

Check schema compatibility before serving traffic on a new version.

package main
 
import (
    "database/sql"
    "errors"
)
 
func assertColumnExists(db *sql.DB, table, column string) error {
    var exists bool
    q := `SELECT EXISTS (
        SELECT 1 FROM information_schema.columns
        WHERE table_name = $1 AND column_name = $2
    )`
    if err := db.QueryRow(q, table, column).Scan(&exists); err != nil {
        return err
    }
    if !exists {
        return errors.New("missing required column: " + table + "." + column)
    }
    return nil
}

Call from /readyz when a new version depends on an expanded column.
Run expand migrations before rolling out binaries that read the column.
Contract (drop old column) only after 100% of pods run the new code.

Related: Rollback Runbooks for Go Deployments - schema rollback steps

Stack versions: This page was written for Go 1.26.x (Green Tea GC default, go fix modernizers - verify patch at build), chi (latest - verify at build), gin (latest - verify at build), echo (latest - verify at build), google.golang.org/grpc (latest - verify at build), sigs.k8s.io/controller-runtime (latest - verify at build), kubebuilder (latest - verify at build), tinygo (latest - verify board targets at build), wazero (latest - verify at build), and golangci-lint (latest - verify linter set at build).

Enterprise Delivery Basics

10 examples to get you started with Enterprise Delivery - 7 basic and 3 intermediate.

Prerequisites

Install Go 1.26.x or later from go.dev/dl.
Create a module: mkdir deliverylab && cd deliverylab && go mod init example.com/deliverylab.
Optional for flag and metrics examples: go get github.com/open-feature/go-sdk/openfeature github.com/prometheus/client_golang/prometheus.

Basic Examples

1. Embed Build Version at Link Time

Stamp git SHA into the binary for deploy correlation and /healthz responses.

package main
 
import (
    "fmt"
    "runtime/debug"
)
 
func main() {
    fmt.Println("version:", buildVersion())
}
 
func buildVersion() string {
    info, ok := debug.ReadBuildInfo()
    if !ok {
        return "dev"
    }
    for _, s := range info.Settings {
        if s.Key == "vcs.revision" {
            if len(s.Value) >= 7 {
                return s.Value[:7]
            }
            return s.Value
        }
    }
    return "unknown"
}

debug.ReadBuildInfo works when built with module info (go build from a git repo).
Pass -ldflags "-X main.version=..." for explicit stamping in CI.
Log and expose this value on every deploy for rollback decisions.

Related: Shipping Go Services Safely at Enterprise Scale - delivery overview

2. CI Test Gate Script

A minimal pipeline step that fails the build on test or race failures.

#!/usr/bin/env bash
set -euo pipefail
go version
go test ./... -count=1 -race -timeout 5m
go vet ./...

-race catches data races before prod - worth the CI time on service code.
-count=1 disables test cache so CI always re-runs.
Wrap this in GitHub Actions, GitLab CI, or Buildkite before image build.

Related: DORA Metrics for Go Teams - measuring pipeline health

3. Multi-Stage Dockerfile for Reproducible Images

Compile in a builder stage; ship a minimal runtime image.

FROM golang:1.26-bookworm AS builder
WORKDIR /src
COPY go.mod go.sum ./
RUN go mod download
COPY . .
RUN CGO_ENABLED=0 GOOS=linux go build -trimpath -ldflags="-s -w" -o /out/app ./cmd/api
 
FROM gcr.io/distroless/static-debian12
COPY --from=builder /out/app /app
USER nonroot:nonroot
ENTRYPOINT ["/app"]

CGO_ENABLED=0 yields a static binary suitable for distroless.
-trimpath removes local paths from binaries for reproducibility.
Tag the built image with git SHA, not :latest, in your registry.

Related: Release Strategy: Blue-Green & Canary for Go - deploying images

4. Readiness Probe Handler

Separate cheap liveness from dependency-checked readiness.

package main
 
import (
    "database/sql"
    "net/http"
    "os"
 
    _ "github.com/jackc/pgx/v5/stdlib"
)
 
func main() {
    db, _ := sql.Open("pgx", os.Getenv("DATABASE_URL"))
    mux := http.NewServeMux()
    mux.HandleFunc("GET /healthz", func(w http.ResponseWriter, _ *http.Request) {
        w.WriteHeader(http.StatusOK)
    })
    mux.HandleFunc("GET /readyz", func(w http.ResponseWriter, _ *http.Request) {
        if err := db.Ping(); err != nil {
            http.Error(w, "db down", http.StatusServiceUnavailable)
            return
        }
        w.WriteHeader(http.StatusOK)
    })
    http.ListenAndServe(":8080", mux)
}

Liveness should not call external systems - only readiness should.
Return 503 on readiness failure so load balancers stop traffic during rollout.
Kubernetes uses these paths in livenessProbe and readinessProbe.

Related: Progressive Delivery & Automated Verification - automated checks

5. Graceful Shutdown for Rolling Deploys

Drain in-flight requests before SIGKILL during pod termination.

package main
 
import (
    "context"
    "log/slog"
    "net/http"
    "os"
    "os/signal"
    "syscall"
    "time"
)
 
func main() {
    srv := &http.Server{Addr: ":8080", Handler: http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) {
        time.Sleep(2 * time.Second)
        w.WriteHeader(http.StatusOK)
    })}
    go func() { _ = srv.ListenAndServe() }()
 
    quit := make(chan os.Signal, 1)
    signal.Notify(quit, syscall.SIGTERM, syscall.SIGINT)
    <-quit
 
    ctx, cancel := context.WithTimeout(context.Background(), 15*time.Second)
    defer cancel()
    if err := srv.Shutdown(ctx); err != nil {
        slog.Error("shutdown", "err", err)
    }
}

Kubernetes sends SIGTERM before removing endpoints from Service.
Shutdown waits for active handlers up to the timeout.
Match terminationGracePeriodSeconds to your worst-case request duration.

Related: Rollback Runbooks for Go Deployments - recovery during bad rollouts

6. Environment-Based Config (12-Factor)

Same binary across staging and prod; config from environment only.

package main
 
import (
    "log/slog"
    "os"
)
 
type Config struct {
    Port        string
    DatabaseURL string
    LogLevel    string
}
 
func loadConfig() Config {
    return Config{
        Port:        envOr("PORT", "8080"),
        DatabaseURL: os.Getenv("DATABASE_URL"),
        LogLevel:    envOr("LOG_LEVEL", "info"),
    }
}
 
func envOr(key, def string) string {
    if v := os.Getenv(key); v != "" {
        return v
    }
    return def
}
 
func main() {
    cfg := loadConfig()
    slog.Info("starting", "port", cfg.Port, "log_level", cfg.LogLevel)
}

Never bake environment-specific values into the image at build time.
Validate required vars at startup and fail fast with clear logs.
Secrets belong in a secret manager, referenced as env vars in the pod spec.

Related: Change Management & Compliance Gates - controlled prod changes

7. Post-Deploy Smoke Test

A tiny Go test binary CI runs against staging after deploy.

package main
 
import (
    "fmt"
    "net/http"
    "os"
    "time"
)
 
func main() {
    base := os.Getenv("SMOKE_BASE_URL")
    if base == "" {
        base = "http://localhost:8080"
    }
    client := &http.Client{Timeout: 5 * time.Second}
    for _, path := range []string{"/healthz", "/readyz"} {
        resp, err := client.Get(base + path)
        if err != nil || resp.StatusCode != http.StatusOK {
            fmt.Fprintf(os.Stderr, "smoke failed %s: %v status=%v\n", path, err, resp)
            os.Exit(1)
        }
        resp.Body.Close()
    }
    fmt.Println("smoke ok")
}

Run smoke immediately after promote to staging; block prod if it fails.
Extend with authenticated happy-path API calls for critical flows.
Keep smoke fast (under 30 seconds) so pipelines stay snappy.

Related: Progressive Delivery & Automated Verification - verification depth

Intermediate Examples

8. Feature Flag Interface (Dark Launch)

Deploy new code with behavior off until ops enables the flag.

package main
 
import (
    "context"
    "os"
)
 
type Flags interface {
    Bool(ctx context.Context, key string, defaultVal bool) bool
}
 
type EnvFlags struct{}
 
func (EnvFlags) Bool(_ context.Context, key string, defaultVal bool) bool {
    if v := os.Getenv("FLAG_" + key); v == "true" {
        return true
    }
    if v == "false" {
        return false
    }
    return defaultVal
}
 
func runNewCheckout(ctx context.Context) error    { return nil }
func runLegacyCheckout(ctx context.Context) error { return nil }
 
func handleCheckout(ctx context.Context, flags Flags) error {
    if flags.Bool(ctx, "NEW_CHECKOUT", false) {
        return runNewCheckout(ctx)
    }
    return runLegacyCheckout(ctx)
}

Start with env-backed flags for simplicity; graduate to OpenFeature or LaunchDarkly.
Default new flags to false in production config.
Dark launch lets canary pods run new paths without user exposure.

Related: Feature Flags in Go Services - production flag SDKs

9. Prometheus Deploy Info Metric

Expose version labels so dashboards mark deploy boundaries.

package main
 
import (
    "net/http"
    "runtime"
 
    "github.com/prometheus/client_golang/prometheus"
    "github.com/prometheus/client_golang/prometheus/promhttp"
)
 
var buildInfo = prometheus.NewGaugeVec(
    prometheus.GaugeOpts{Name: "go_build_info", Help: "Build metadata"},
    []string{"version", "goversion"},
)
 
func main() {
    prometheus.MustRegister(buildInfo)
    buildInfo.WithLabelValues("abc1234", runtime.Version()).Set(1)
    http.Handle("/metrics", promhttp.Handler())
    http.ListenAndServe(":9090", nil)
}

Register metrics once in main, never per request.
Grafana annotations can watch go_build_info label changes for deploy lines.
Pair with RED charts to spot CFR spikes after version label changes.

Related: DORA Metrics for Go Teams - change failure rate

10. Expand-Contract Migration Guard

Check schema compatibility before serving traffic on a new version.

package main
 
import (
    "database/sql"
    "errors"
)
 
func assertColumnExists(db *sql.DB, table, column string) error {
    var exists bool
    q := `SELECT EXISTS (
        SELECT 1 FROM information_schema.columns
        WHERE table_name = $1 AND column_name = $2
    )`
    if err := db.QueryRow(q, table, column).Scan(&exists); err != nil {
        return err
    }
    if !exists {
        return errors.New("missing required column: " + table + "." + column)
    }
    return nil
}

Call from /readyz when a new version depends on an expanded column.
Run expand migrations before rolling out binaries that read the column.
Contract (drop old column) only after 100% of pods run the new code.

Related: Rollback Runbooks for Go Deployments - schema rollback steps

Stack versions: This page was written for Go 1.26.x (Green Tea GC default, go fix modernizers - verify patch at build), chi (latest - verify at build), gin (latest - verify at build), echo (latest - verify at build), google.golang.org/grpc (latest - verify at build), sigs.k8s.io/controller-runtime (latest - verify at build), kubebuilder (latest - verify at build), tinygo (latest - verify board targets at build), wazero (latest - verify at build), and golangci-lint (latest - verify linter set at build).

Enterprise Delivery Basics

Search Documentation

Enterprise Delivery Basics