Make keepalive parameters configurable via environment variables

Address review feedback: keepalive timings are now configurable through
environment variables instead of being hardcoded.

Environment variables:
- GNMI_CLIENT_KEEPALIVE_INTERVAL: Client ping interval in seconds (default: 30)
- GNMI_CLIENT_KEEPALIVE_TIMEOUT: Client ping ack timeout in seconds (default: 10)
- GNMI_SERVER_KEEPALIVE_MIN_TIME: Server minimum allowed ping interval in seconds (default: 20)

Signed-off-by: Dawei Huang <daweihuang@microsoft.com>

Author: hdwhdw

pkg/interceptors/dpuproxy/proxy.go

@@ -4,6 +4,8 @@ import (
 	"context"
 	"fmt"
 	"io"
+	"os"
+	"strconv"
 	"sync"
 	"time"
 
@@ -77,6 +79,17 @@ var defaultForwardableMethods = []ForwardableMethod{
 
 var defaultProxy *DPUProxy
 
+// getEnvDuration reads an environment variable as a duration in seconds.
+// Returns the fallback if the variable is unset or invalid.
+func getEnvDuration(key string, fallback time.Duration) time.Duration {
+	if v := os.Getenv(key); v != "" {
+		if secs, err := strconv.Atoi(v); err == nil && secs > 0 {
+			return time.Duration(secs) * time.Second
+		}
+	}
+	return fallback
+}
+
 // SetDefaultProxy registers the DPU proxy singleton for use by handlers
 // that need direct DPU connections (e.g., TransferToRemote file operations).
 func SetDefaultProxy(p *DPUProxy) { defaultProxy = p }
@@ -187,9 +200,9 @@ func (p *DPUProxy) getConnection(ctx context.Context, dpuIndex, ipAddress string
 			target,
 			grpc.WithTransportCredentials(insecure.NewCredentials()),
 			grpc.WithKeepaliveParams(keepalive.ClientParameters{
-				Time:                30 * time.Second, // Send keepalive ping every 30s
-				Timeout:             10 * time.Second, // Wait 10s for ping ack before considering connection dead
-				PermitWithoutStream: true,             // Send pings even when no active RPCs
+				Time:                getEnvDuration("GNMI_CLIENT_KEEPALIVE_INTERVAL", 30*time.Second),
+				Timeout:             getEnvDuration("GNMI_CLIENT_KEEPALIVE_TIMEOUT", 10*time.Second),
+				PermitWithoutStream: true,
 			}),
 		)
 

telemetry/telemetry.go

@@ -12,6 +12,7 @@ import (
 	"os"
 	"os/signal"
 	"path/filepath"
+	"strconv"
 	"strings"
 	"sync"
 	"sync/atomic"
@@ -532,9 +533,15 @@ func startGNMIServer(telemetryCfg *TelemetryConfig, cfg *gnmi.Config, serverCont
 			// causing "too_many_pings" GOAWAY for clients that ping more
 			// frequently during long-running operations like SetPackage.
 			// See: https://github.com/sonic-net/sonic-gnmi/issues/619
+			serverMinPingTime := 20 * time.Second
+			if v := os.Getenv("GNMI_SERVER_KEEPALIVE_MIN_TIME"); v != "" {
+				if secs, err := strconv.Atoi(v); err == nil && secs > 0 {
+					serverMinPingTime = time.Duration(secs) * time.Second
+				}
+			}
 			keep_alive_policy := keepalive.EnforcementPolicy{
-				MinTime:             20 * time.Second, // Allow pings as frequent as every 20s
-				PermitWithoutStream: true,             // Allow pings when there are no active streams
+				MinTime:             serverMinPingTime,
+				PermitWithoutStream: true,
 			}
 			commonOpts = append(commonOpts, grpc.KeepaliveEnforcementPolicy(keep_alive_policy))