change lib

INSTALL.md

@@ -25,6 +25,12 @@ As root:
 go get -v -u github.com/FreifunkBremen/yanic
 ```
 
+### allow to ping
+only needed if config has `nodes.ping_count` > 0
+```sh
+sudo setcap cap_net_raw=+ep  /opt/go/bin/yanic
+```
+
 #### Work with other databases
 If you like to use another database solution than influxdb, Pull Requests are
 welcome. Just fork this project and create another subpackage within the folder

config_example.toml

@@ -55,6 +55,12 @@ save_interval = "5s"
 # Set node to offline if not seen within this period
 offline_after = "10m"
 
+## Verify if node is really down by ping last seen address of node
+# send x pings to verify if node is offline (for disable set count < 1)
+ping_count = 3
+# timeout of sending ping to a node
+ping_timeout = "1s"
+
 
 ## [[nodes.output.example]]
 # Each output format has its own config block and needs to be enabled by adding:

database/graphite/global.go

@@ -36,6 +36,7 @@ func (c *Connection) InsertGlobals(stats *runtime.GlobalStats, time time.Time, s
 func GlobalStatsFields(name string, stats *runtime.GlobalStats) []graphigo.Metric {
 	return []graphigo.Metric{
 		{Name: name + ".nodes", Value: stats.Nodes},
+		{Name: name + ".nodes.no_respondd", Value: stats.NodesNoRespondd},
 		{Name: name + ".gateways", Value: stats.Gateways},
 		{Name: name + ".clients.total", Value: stats.Clients},
 		{Name: name + ".clients.wifi", Value: stats.ClientsWifi},

database/influxdb/global.go

@@ -43,6 +43,7 @@ func (conn *Connection) InsertGlobals(stats *runtime.GlobalStats, time time.Time
 func GlobalStatsFields(stats *runtime.GlobalStats) map[string]interface{} {
 	return map[string]interface{}{
 		"nodes":             stats.Nodes,
+		"nodes.no_respondd": stats.NodesNoRespondd,
 		"gateways":          stats.Gateways,
 		"clients.total":     stats.Clients,
 		"clients.wifi":      stats.ClientsWifi,

database/logging/file.go

@@ -50,7 +50,7 @@ func (conn *Connection) InsertLink(link *runtime.Link, time time.Time) {
 }
 
 func (conn *Connection) InsertGlobals(stats *runtime.GlobalStats, time time.Time, site string, domain string) {
-	conn.log("InsertGlobals: [", time.String(), "] site: ", site, " domain: ", domain, ", nodes: ", stats.Nodes, ", clients: ", stats.Clients, " models: ", len(stats.Models))
+	conn.log("InsertGlobals: [", time.String(), "] site: ", site, " domain: ", domain, ", nodes: ", stats.Nodes, " (no respondd: ", stats.NodesNoRespondd, "), clients: ", stats.Clients, " models: ", len(stats.Models))
 }
 
 func (conn *Connection) PruneNodes(deleteAfter time.Duration) {

docs/docs_configuration.md

@@ -203,6 +203,8 @@ state_path     = "/var/lib/yanic/state.json"
 prune_after    = "7d"
 save_interval  = "5s"
 offline_after  = "10m"
+ping_count = 3
+ping_timeout = "1s"
 ```
 {% endmethod %}
 
@@ -246,6 +248,26 @@ offline_after = "10m"
 ```
 {% endmethod %}
 
+### ping_count
+{% method %}
+Verify if node is really down by ping last seen address of node
+send x pings to verify if node is offline (for disable set count < 1)
+{% sample lang="toml" %}
+```toml
+ping_count = 3
+```
+{% endmethod %}
+
+
+### ping_timeout
+{% method %}
+Timeout of sending ping to a node
+{% sample lang="toml" %}
+```toml
+ping_timeout = "1s"
+```
+{% endmethod %}
+
 
 ## [[nodes.output.example]]
 {% method %}

docs/docs_install.md

@@ -26,6 +26,12 @@ As root:
 go get -v -u github.com/FreifunkBremen/yanic
 ```
 
+### allow to ping
+only needed if config has `nodes.ping_count` > 0
+```sh
+sudo setcap cap_net_raw=+ep  /opt/go/bin/yanic
+```
+
 ### Install
 
 ```sh

runtime/node.go

@@ -13,6 +13,7 @@ type Node struct {
 	Firstseen  jsontime.Time    `json:"firstseen"`
 	Lastseen   jsontime.Time    `json:"lastseen"`
 	Online     bool             `json:"online"`
+	NoRespondd bool             `json:"no_respondd"`
 	Statistics *data.Statistics `json:"statistics"`
 	Nodeinfo   *data.NodeInfo   `json:"nodeinfo"`
 	Neighbours *data.Neighbours `json:"-"`

runtime/nodes.go

@@ -7,6 +7,7 @@ import (
 	"time"
 
 	"github.com/bdlm/log"
+	ping "github.com/digineo/go-ping"
 
 	"github.com/FreifunkBremen/yanic/data"
 	"github.com/FreifunkBremen/yanic/lib/jsontime"
@@ -17,6 +18,7 @@ type Nodes struct {
 	List          map[string]*Node  `json:"nodes"` // the current nodemap, indexed by node ID
 	ifaceToNodeID map[string]string // mapping from MAC address to NodeID
 	config        *NodesConfig
+	pinger          *ping.Pinger
 	sync.RWMutex
 }
 
@@ -27,6 +29,11 @@ func NewNodes(config *NodesConfig) *Nodes {
 		ifaceToNodeID: make(map[string]string),
 		config:        config,
 	}
+	p, err := ping.New("", "::")
+	if err != nil {
+		log.Warnf("ping bind failed: %s", err)
+	}
+	nodes.pinger = p
 
 	if config.StatePath != "" {
 		nodes.load()
@@ -80,9 +87,10 @@ func (nodes *Nodes) Update(nodeID string, res *data.ResponseData) *Node {
 	// Update fields
 	node.Lastseen = now
 	node.Online = true
-	node.Neighbours = res.Neighbours
 	node.Nodeinfo = res.NodeInfo
 	node.Statistics = res.Statistics
+	node.Neighbours = res.Neighbours
+	node.NoRespondd = false
 
 	return node
 }
@@ -170,15 +178,29 @@ func (nodes *Nodes) expire() {
 	nodes.Lock()
 	defer nodes.Unlock()
 
+	wg := sync.WaitGroup{}
+
 	for id, node := range nodes.List {
 		if node.Lastseen.Before(pruneAfter) {
 			// expire
 			delete(nodes.List, id)
 		} else if node.Lastseen.Before(offlineAfter) {
 			// set to offline
+			wg.Add(1)
+			go func(node *Node) {
+				defer wg.Done()
+				if nodes.config.PingCount > 0 && nodes.ping(node) {
+					node.Online = true
+					node.NoRespondd = true
+				} else {
 					node.Online = false
+					node.NoRespondd = false
+				}
+			}(node)
 		}
 	}
+	wg.Wait()
+	log.WithField("nodes", "expire").Info("end")
 }
 
 // adds the nodes interface addresses to the internal map
@@ -240,6 +262,7 @@ func (nodes *Nodes) save() {
 
 	// serialize nodes
 	SaveJSON(nodes, nodes.config.StatePath)
+	log.WithField("nodes", "save").Info("end")
 }
 
 // SaveJSON to path

runtime/nodes_config.go

@@ -7,5 +7,7 @@ type NodesConfig struct {
 	SaveInterval duration.Duration `toml:"save_interval"` // Save nodes periodically
 	OfflineAfter duration.Duration `toml:"offline_after"` // Set node to offline if not seen within this period
 	PruneAfter   duration.Duration `toml:"prune_after"`   // Remove nodes after n days of inactivity
+	PingCount    int               `toml:"ping_count"`    // send x pings to verify if node is offline (for disable count < 1)
+	PingTimeout  duration.Duration `toml:"ping_timeout"`  // timeout of sending ping to a node
 	Output       map[string]interface{}
 }

runtime/nodes_ping.go

@@ -0,0 +1,39 @@
+package runtime
+
+import (
+	"net"
+
+	"github.com/bdlm/log"
+)
+
+func (nodes *Nodes) ping(node *Node) bool {
+	logNode := log.WithField("node_id", "unknown")
+	if node.Nodeinfo != nil {
+		logNode = logNode.WithField("node_id", node.Nodeinfo.NodeID)
+	}
+	var addr *net.IPAddr
+	if node.Address != nil {
+		addr = &net.IPAddr{IP:node.Address.IP, Zone: node.Address.Zone}
+	} else {
+		logNode.Debug("error no address found")
+		if node.Nodeinfo != nil {
+			for _, addrMaybeString := range node.Nodeinfo.Network.Addresses {
+				if len(addrMaybeString) >= 5 && addrMaybeString[:5] != "fe80:" {
+					addrMaybe, err := net.ResolveIPAddr("ip6", addrMaybeString)
+					if err == nil {
+						addr = addrMaybe
+					}
+				}
+			}
+		}
+	}
+
+	logAddr := logNode.WithField("addr", addr.String())
+
+	_, err := nodes.pinger.PingAttempts(addr, nodes.config.PingTimeout.Duration, nodes.config.PingCount)
+
+	logAddr.WithFields(map[string]interface{}{
+		"success": err == nil,
+	}).Debug("pong")
+	return err == nil
+}

runtime/nodes_ping_test.go

@@ -0,0 +1,44 @@
+package runtime
+
+import (
+	"net"
+	"testing"
+	"time"
+
+	"github.com/bdlm/log"
+	"github.com/stretchr/testify/assert"
+
+	"github.com/FreifunkBremen/yanic/data"
+)
+
+func TestPing(t *testing.T) {
+	log.SetLevel(log.DebugLevel)
+
+	assert := assert.New(t)
+	config := &NodesConfig{
+		PingCount: 1,
+	}
+	config.OfflineAfter.Duration = time.Minute * 10
+	// to get default (100%) path of testing
+	// config.PruneAfter.Duration = time.Hour * 24 * 6
+	nodes := &Nodes{
+		config:        config,
+		List:          make(map[string]*Node),
+		ifaceToNodeID: make(map[string]string),
+	}
+
+	node := nodes.Update("expire", &data.ResponseData{NodeInfo: &data.NodeInfo{
+		NodeID:  "nodeID-Lola",
+		Network: data.Network{Addresses: []string{"fe80::1", "fd2f::1"}},
+	}})
+	// get fallback
+	assert.False(nodes.ping(node))
+
+	node.Address = &net.UDPAddr{Zone: "bat0"}
+	// error during ping
+	assert.False(nodes.ping(node))
+
+	node.Address.IP = net.ParseIP("fe80::1")
+	// error during ping
+	assert.False(nodes.ping(node))
+}

runtime/stats.go

@@ -17,6 +17,7 @@ type GlobalStats struct {
 	ClientsWifi5    uint32
 	Gateways        uint32
 	Nodes           uint32
+	NodesNoRespondd uint32
 
 	Firmwares   CounterMap
 	Models      CounterMap
@@ -81,6 +82,9 @@ func (s *GlobalStats) Add(node *Node) {
 		s.ClientsWifi5 += stats.Clients.Wifi5
 		s.ClientsWifi += stats.Clients.Wifi
 	}
+	if node.NoRespondd {
+		s.NodesNoRespondd++
+	}
 	if node.IsGateway() {
 		s.Gateways++
 	}

runtime/stats_test.go

@@ -22,6 +22,7 @@ func TestGlobalStats(t *testing.T) {
 	//check GLOBAL_SITE stats
 	assert.EqualValues(1, stats[GLOBAL_SITE][GLOBAL_DOMAIN].Gateways)
 	assert.EqualValues(3, stats[GLOBAL_SITE][GLOBAL_DOMAIN].Nodes)
+	assert.EqualValues(1, stats[GLOBAL_SITE][GLOBAL_DOMAIN].NodesNoRespondd)
 	assert.EqualValues(25, stats[GLOBAL_SITE][GLOBAL_DOMAIN].Clients)
 
 	// check models
@@ -99,6 +100,7 @@ func createTestNodes() *Nodes {
 
 	nodes.AddNode(&Node{
 		Online:     true,
+		NoRespondd: true,
 		Statistics: &data.Statistics{
 			Clients: data.Clients{
 				Total: 2,