session: allow log configuration using Logger interface.

The driver now provides a Logger interface allowing to supply
a custom logger as part of (Session)ConnConfig.

Driver also provides three implementations of Logger by itself:
- DefaultLogger, logging only warnings
- DebugLogger, logging everything that can be useful for debugging,
  it is used by default in tests
- NopLogger, logging nothing

Fixes #271

log/logger.go

@@ -0,0 +1,69 @@
+package log
+
+import (
+	"log"
+	"os"
+)
+
+type Logger interface {
+	Info(v ...any)
+	Infof(format string, v ...any)
+	Infoln(v ...any)
+
+	Warn(v ...any)
+	Warnf(format string, v ...any)
+	Warnln(v ...any)
+}
+
+// DefaultLogger only logs warnings and critical errors.
+type DefaultLogger struct {
+	warn *log.Logger
+}
+
+func NewDefaultLogger() *DefaultLogger {
+	res := &DefaultLogger{
+		warn: log.New(os.Stderr, "WARNING ", log.LstdFlags),
+	}
+	return res
+}
+
+func (logger *DefaultLogger) Info(v ...any)                 {}
+func (logger *DefaultLogger) Infof(format string, v ...any) {}
+func (logger *DefaultLogger) Infoln(v ...any)               {}
+
+func (logger *DefaultLogger) Warn(v ...any)                 { logger.warn.Print(v...) }
+func (logger *DefaultLogger) Warnf(format string, v ...any) { logger.warn.Printf(format, v...) }
+func (logger *DefaultLogger) Warnln(v ...any)               { logger.warn.Println(v...) }
+
+// DebugLogger logs both warnings and information about important events in driver's runtime.
+type DebugLogger struct {
+	info *log.Logger
+	warn *log.Logger
+}
+
+func NewDebugLogger() *DebugLogger {
+	res := &DebugLogger{
+		info: log.New(os.Stderr, "INFO ", log.LstdFlags),
+		warn: log.New(os.Stderr, "WARNING ", log.LstdFlags),
+	}
+	return res
+}
+
+func (logger *DebugLogger) Info(v ...any)                 { logger.info.Print(v...) }
+func (logger *DebugLogger) Infof(format string, v ...any) { logger.info.Printf(format, v...) }
+func (logger *DebugLogger) Infoln(v ...any)               { logger.info.Println(v...) }
+
+func (logger *DebugLogger) Warn(v ...any)                 { logger.warn.Print(v...) }
+func (logger *DebugLogger) Warnf(format string, v ...any) { logger.warn.Printf(format, v...) }
+func (logger *DebugLogger) Warnln(v ...any)               { logger.warn.Println(v...) }
+
+// NopLogger doesn't log anything.
+type NopLogger struct{}
+
+func (NopLogger) Info(v ...any)                 {}
+func (NopLogger) Infof(format string, v ...any) {}
+func (NopLogger) Infoln(v ...any)               {}
+
+func (NopLogger) Warn(v ...any)                 {}
+func (NopLogger) Warnf(format string, v ...any) {}
+func (NopLogger) Warnln(v ...any)               {}

session.go

@@ -3,7 +3,6 @@ package scylla
 import (
 	"context"
 	"fmt"
-	"log"
 	"sync"
 	"time"
 
@@ -271,6 +270,6 @@ func (s *Session) NewTokenAwareDCAwarePolicy(localDC string) transport.HostSelec
 }
 
 func (s *Session) Close() {
-	log.Println("session: close")
+	s.cfg.Logger.Info("session: close")
 	s.cluster.Close()
 }

session_integration_test.go

@@ -18,13 +18,18 @@ import (
 
 	"github.com/scylladb/scylla-go-driver/frame"
 	"github.com/scylladb/scylla-go-driver/frame/response"
+	"github.com/scylladb/scylla-go-driver/log"
 	"github.com/scylladb/scylla-go-driver/transport"
 	"go.uber.org/goleak"
 )
 
 const TestHost = "192.168.100.100"
 
-var testingSessionConfig = DefaultSessionConfig("mykeyspace", TestHost)
+var testingSessionConfig = func() SessionConfig {
+	cfg := DefaultSessionConfig("mykeyspace", TestHost)
+	cfg.Logger = log.NewDebugLogger()
+	return cfg
+}()
 
 func initKeyspace(ctx context.Context, t testing.TB) {
 	t.Helper()

transport/cluster.go

@@ -3,7 +3,6 @@ package transport
 import (
 	"context"
 	"fmt"
-	"log"
 	"net"
 	"sort"
 	"strconv"
@@ -160,7 +159,7 @@ func NewCluster(ctx context.Context, cfg ConnConfig, p HostSelectionPolicy, e []
 }
 
 func (c *Cluster) NewControl(ctx context.Context) (*Conn, error) {
-	log.Printf("cluster: open control connection")
+	c.cfg.Logger.Info("cluster: open control connection")
 	var errs []string
 	for addr := range c.knownHosts {
 		conn, err := OpenConn(ctx, addr, nil, c.cfg)
@@ -184,7 +183,7 @@ func (c *Cluster) NewControl(ctx context.Context) (*Conn, error) {
 // refreshTopology creates new topology filled with the result of keyspaceQuery, localQuery and peerQuery.
 // Old topology is replaced with the new one atomically to prevent dirty reads.
 func (c *Cluster) refreshTopology(ctx context.Context) error {
-	log.Printf("cluster: refresh topology")
+	c.cfg.Logger.Infoln("cluster: refresh topology")
 	rows, err := c.getAllNodesInfo(ctx)
 	if err != nil {
 		return fmt.Errorf("query info about nodes in cluster: %w", err)
@@ -236,9 +235,9 @@ func (c *Cluster) refreshTopology(ctx context.Context) error {
 	}
 
 	if ks, ok := t.keyspaces[c.cfg.Keyspace]; ok {
-		t.policyInfo.Preprocess(t, ks)
+		t.policyInfo.Preprocess(t, ks, c.cfg.Logger)
 	} else {
-		t.policyInfo.Preprocess(t, keyspace{})
+		t.policyInfo.Preprocess(t, keyspace{}, c.cfg.Logger)
 	}
 
 	c.setTopology(t)
@@ -450,7 +449,7 @@ func (c *Cluster) setTopology(t *topology) {
 // of registering handlers for them.
 func (c *Cluster) handleEvent(ctx context.Context, r response) {
 	if r.Err != nil {
-		log.Printf("cluster: received event with error: %v", r.Err)
+		c.cfg.Logger.Infoln("cluster: received event with error: %v", r.Err)
 		c.RequestReopenControl()
 		return
 	}
@@ -462,17 +461,17 @@ func (c *Cluster) handleEvent(ctx context.Context, r response) {
 	case *SchemaChange:
 		// TODO: add schema change.
 	default:
-		log.Printf("cluster: unsupported event type: %v", r.Response)
+		c.cfg.Logger.Warnf("cluster: unsupported event type: %v", r.Response)
 	}
 }
 
 func (c *Cluster) handleTopologyChange(v *TopologyChange) {
-	log.Printf("cluster: handle topology change: %+#v", v)
+	c.cfg.Logger.Infof("cluster: handle topology change: %+#v", v)
 	c.RequestRefresh()
 }
 
 func (c *Cluster) handleStatusChange(ctx context.Context, v *StatusChange) {
-	log.Printf("cluster: handle status change: %+#v", v)
+	c.cfg.Logger.Infof("cluster: handle status change: %+#v", v)
 	m := c.Topology().peers
 	addr := v.Address.String()
 	if n, ok := m[addr]; ok {
@@ -482,10 +481,10 @@ func (c *Cluster) handleStatusChange(ctx context.Context, v *StatusChange) {
 		case frame.Down:
 			n.setStatus(statusDown)
 		default:
-			log.Printf("cluster: status change not supported: %+#v", v)
+			c.cfg.Logger.Warnf("cluster: status change not supported: %+#v", v)
 		}
 	} else {
-		log.Printf("cluster: unknown node %s received status change: %+#v in topology %v", addr, v, m)
+		c.cfg.Logger.Infof("cluster: unknown node %s received status change: %+#v in topology %v, requesting topology refresh", addr, v, m)
 		c.RequestRefresh()
 	}
 }
@@ -503,7 +502,7 @@ func (c *Cluster) loop(ctx context.Context) {
 		case <-c.reopenControlChan:
 			c.tryReopenControl(ctx)
 		case <-ctx.Done():
-			log.Printf("cluster closing due to: %v", ctx.Err())
+			c.cfg.Logger.Infof("cluster closing due to: %v", ctx.Err())
 			c.handleClose()
 			return
 		case <-c.closeChan:
@@ -523,17 +522,17 @@ func (c *Cluster) tryRefresh(ctx context.Context) {
 	if err := c.refreshTopology(ctx); err != nil {
 		c.RequestReopenControl()
 		time.AfterFunc(tryRefreshInterval, c.RequestRefresh)
-		log.Printf("cluster: refresh topology: %v", err)
+		c.cfg.Logger.Infof("cluster: refresh topology: %v", err)
 	}
 }
 
 const tryReopenControlInterval = time.Second
 
 func (c *Cluster) tryReopenControl(ctx context.Context) {
-	log.Printf("cluster: reopen control connection")
+	c.cfg.Logger.Infoln("cluster: reopen control connection")
 	if control, err := c.NewControl(ctx); err != nil {
 		time.AfterFunc(tryReopenControlInterval, c.RequestReopenControl)
-		log.Printf("cluster: failed to reopen control connection: %v", err)
+		c.cfg.Logger.Infof("cluster: failed to reopen control connection: %v", err)
 	} else {
 		c.control.Close()
 		c.control = control
@@ -542,7 +541,7 @@ func (c *Cluster) tryReopenControl(ctx context.Context) {
 }
 
 func (c *Cluster) handleClose() {
-	log.Printf("cluster: handle cluster close")
+	c.cfg.Logger.Infoln("cluster: handle cluster close")
 	c.control.Close()
 	m := c.Topology().peers
 	for _, n := range m {
@@ -551,23 +550,23 @@ func (c *Cluster) handleClose() {
 }
 
 func (c *Cluster) RequestRefresh() {
-	log.Printf("cluster: requested to refresh cluster topology")
+	c.cfg.Logger.Infoln("cluster: requested to refresh cluster topology")
 	select {
 	case c.refreshChan <- struct{}{}:
 	default:
 	}
 }
 
 func (c *Cluster) RequestReopenControl() {
-	log.Printf("cluster: requested to reopen control connection")
+	c.cfg.Logger.Infoln("cluster: requested to reopen control connection")
 	select {
 	case c.reopenControlChan <- struct{}{}:
 	default:
 	}
 }
 
 func (c *Cluster) Close() {
-	log.Printf("cluster: requested to close cluster")
+	c.cfg.Logger.Infoln("cluster: requested to close cluster")
 	select {
 	case c.closeChan <- struct{}{}:
 	default: