grafana-iam: Adds config opts for RBACRemoteClient for load balancing (#110819)

pkg/services/grpcserver/README.md

@@ -12,16 +12,16 @@ The `grpcserver` package provides the implementation of the gRPC server for hand
 
 ## Usage
 
-Enable the gRPC server in Grafana by setting the `grpcServer` feature toggle to `true` in your `custom.ini` configuration file. 
+Enable the gRPC server in Grafana by setting the `grpcServer` feature toggle to `true` in your `custom.ini` configuration file.
 
-``` ini
+```ini
 [feature_toggles]
 grpcServer = true
 ```
 
 You can specify the gRPC server specific settings in the `grpc_server` section of the configuration file.
 
-``` ini
+```ini
 [grpc_server]
 network = "tcp"
 address = "127.0.0.1:10000"
@@ -36,6 +36,31 @@ max_recv_msg_size =
 max_send_msg_size =
 ```
 
+### Optional: Connection Management and Load Balancing
+
+These settings help with:
+
+- **Resource management**: Prevent resource leaks from idle connections
+- **Connection health**: Detect and clean up dead connections
+- **Load balancing**: Force connection recycling for better distribution across multiple server instances
+- **DoS protection**: Rate limit keepalive pings from clients
+
+```ini
+# Connection management options
+# Maximum amount of time a connection may exist before it will be closed
+max_connection_age = 300s
+# Additional time to allow for pending RPCs to complete before forcibly closing connections
+max_connection_age_grace = 10s
+# Maximum amount of idle time before a connection is closed
+max_connection_idle = 300s
+# Frequency of server-to-client pings to check if a connection is still active
+keepalive_time = 30s
+# Amount of time the server waits for a response to keepalive pings before closing the connection
+keepalive_timeout = 5s
+# Minimum amount of time a client should wait before sending a keepalive ping
+keepalive_min_time = 5s
+```
+
 ## Example Services
 
-View [health.go] and [reflection.go] for examples of how to implement gRPC services in Grafana. These services are currently initialized by the [background service registry](../../registry/backgroundsvcs/background_services.go).
+View [health.go] and [reflection.go] for examples of how to implement gRPC services in Grafana. These services are currently initialized by the [background service registry](../../registry/backgroundsvcs/background_services.go).

pkg/services/grpcserver/service.go

@@ -14,6 +14,7 @@ import (
 	"go.opentelemetry.io/otel/trace"
 	"google.golang.org/grpc"
 	"google.golang.org/grpc/credentials"
+	"google.golang.org/grpc/keepalive"
 
 	"github.com/grafana/grafana/pkg/infra/log"
 	"github.com/grafana/grafana/pkg/registry"
@@ -104,12 +105,43 @@ func ProvideService(cfg *setting.Cfg, features featuremgmt.FeatureToggles, authe
 		opts = append(opts, grpc.MaxSendMsgSize(s.cfg.MaxSendMsgSize))
 	}
 
+	// Apply connection management settings
+	keepaliveParams := keepalive.ServerParameters{
+		MaxConnectionAge:      s.cfg.MaxConnectionAge,
+		MaxConnectionAgeGrace: s.cfg.MaxConnectionAgeGrace,
+		MaxConnectionIdle:     s.cfg.MaxConnectionIdle,
+		Time:                  s.cfg.KeepaliveTime,
+		Timeout:               s.cfg.KeepaliveTimeout,
+	}
+	keepalivePolicy := keepalive.EnforcementPolicy{
+		MinTime: s.cfg.KeepaliveMinTime,
+	}
+
+	// Only add keepalive options if any values are configured
+	if s.cfg.MaxConnectionAge > 0 || s.cfg.MaxConnectionAgeGrace > 0 || s.cfg.MaxConnectionIdle > 0 ||
+		s.cfg.KeepaliveTime > 0 || s.cfg.KeepaliveTimeout > 0 {
+		opts = append(opts, grpc.KeepaliveParams(keepaliveParams))
+	}
+	if s.cfg.KeepaliveMinTime > 0 {
+		opts = append(opts, grpc.KeepaliveEnforcementPolicy(keepalivePolicy))
+	}
+
 	s.server = grpc.NewServer(opts...)
 	return s, nil
 }
 
 func (s *gPRCServerService) Run(ctx context.Context) error {
-	s.logger.Info("Running GRPC server", "address", s.cfg.Address, "network", s.cfg.Network, "tls", s.cfg.TLSConfig != nil, "max_recv_msg_size", s.cfg.MaxRecvMsgSize, "max_send_msg_size", s.cfg.MaxSendMsgSize)
+	s.logger.Info("Running GRPC server",
+		"address", s.cfg.Address,
+		"network", s.cfg.Network,
+		"tls", s.cfg.TLSConfig != nil,
+		"max_recv_msg_size", s.cfg.MaxRecvMsgSize,
+		"max_send_msg_size", s.cfg.MaxSendMsgSize,
+		"max_connection_age", s.cfg.MaxConnectionAge,
+		"max_connection_idle", s.cfg.MaxConnectionIdle,
+		"keepalive_time", s.cfg.KeepaliveTime,
+		"keepalive_timeout", s.cfg.KeepaliveTimeout,
+		"keepalive_min_time", s.cfg.KeepaliveMinTime)
 
 	listener, err := net.Listen(s.cfg.Network, s.cfg.Address)
 	if err != nil {