add keep-alive support to socket_listener & socket_writer (#2697)

closes #2635

plugins/outputs/socket_writer/README.md

@@ -19,6 +19,12 @@ It can output data in any of the [supported output formats](https://github.com/i
   # address = "unix:///tmp/telegraf.sock"
   # address = "unixgram:///tmp/telegraf.sock"
 
+  ## Period between keep alive probes.
+  ## Only applies to TCP sockets.
+  ## 0 disables keep alive probes.
+  ## Defaults to the OS configuration.
+  # keep_alive_period = "5m"
+
   ## Data format to generate.
   ## Each data format has it's own unique set of configuration options, read
   ## more about them here:

plugins/outputs/socket_writer/socket_writer.go

@@ -2,16 +2,19 @@ package socket_writer
 
 import (
 	"fmt"
+	"log"
 	"net"
 	"strings"
 
 	"github.com/influxdata/telegraf"
+	"github.com/influxdata/telegraf/internal"
 	"github.com/influxdata/telegraf/plugins/outputs"
 	"github.com/influxdata/telegraf/plugins/serializers"
 )
 
 type SocketWriter struct {
-	Address string
+	Address         string
+	KeepAlivePeriod *internal.Duration
 
 	serializers.Serializer
 
@@ -36,6 +39,12 @@ func (sw *SocketWriter) SampleConfig() string {
   # address = "unix:///tmp/telegraf.sock"
   # address = "unixgram:///tmp/telegraf.sock"
 
+  ## Period between keep alive probes.
+  ## Only applies to TCP sockets.
+  ## 0 disables keep alive probes.
+  ## Defaults to the OS configuration.
+  # keep_alive_period = "5m"
+
   ## Data format to generate.
   ## Each data format has it's own unique set of configuration options, read
   ## more about them here:
@@ -59,10 +68,31 @@ func (sw *SocketWriter) Connect() error {
 		return err
 	}
 
+	if err := sw.setKeepAlive(c); err != nil {
+		log.Printf("unable to configure keep alive (%s): %s", sw.Address, err)
+	}
+
 	sw.Conn = c
 	return nil
 }
 
+func (sw *SocketWriter) setKeepAlive(c net.Conn) error {
+	if sw.KeepAlivePeriod == nil {
+		return nil
+	}
+	tcpc, ok := c.(*net.TCPConn)
+	if !ok {
+		return fmt.Errorf("cannot set keep alive on a %s socket", strings.SplitN(sw.Address, "://", 2)[0])
+	}
+	if sw.KeepAlivePeriod.Duration == 0 {
+		return tcpc.SetKeepAlive(false)
+	}
+	if err := tcpc.SetKeepAlive(true); err != nil {
+		return err
+	}
+	return tcpc.SetKeepAlivePeriod(sw.KeepAlivePeriod.Duration)
+}
+
 // Write writes the given metrics to the destination.
 // If an error is encountered, it is up to the caller to retry the same write again later.
 // Not parallel safe.