Dedicated whitelist metrics (#2813)

* add proper whitelist metrics : both its own table and an extension to acquis metrics to track discarded/whitelisted lines

.golangci.yml

@@ -11,7 +11,7 @@ run:
 linters-settings:
   cyclop:
     # lower this after refactoring
-    max-complexity: 66
+    max-complexity: 70
 
   gci:
     sections:
@@ -22,11 +22,11 @@ linters-settings:
 
   gocognit:
     # lower this after refactoring
-    min-complexity: 145
+    min-complexity: 150
 
   gocyclo:
     # lower this after refactoring
-    min-complexity: 64
+    min-complexity: 70
 
   funlen:
     # Checks the number of lines in a function.

cmd/crowdsec-cli/metrics.go

@@ -21,21 +21,22 @@ import (
 )
 
 type (
-	statAcquis map[string]map[string]int
-	statParser map[string]map[string]int
-	statBucket map[string]map[string]int
-	statLapi map[string]map[string]int
-	statLapiMachine map[string]map[string]map[string]int
-	statLapiBouncer map[string]map[string]map[string]int
+	statAcquis       map[string]map[string]int
+	statParser       map[string]map[string]int
+	statBucket       map[string]map[string]int
+	statWhitelist    map[string]map[string]map[string]int
+	statLapi         map[string]map[string]int
+	statLapiMachine  map[string]map[string]map[string]int
+	statLapiBouncer  map[string]map[string]map[string]int
 	statLapiDecision map[string]struct {
 		NonEmpty int
 		Empty    int
 	}
-	statDecision map[string]map[string]map[string]int
+	statDecision     map[string]map[string]map[string]int
 	statAppsecEngine map[string]map[string]int
-	statAppsecRule map[string]map[string]map[string]int
-	statAlert map[string]int
-	statStash map[string]struct {
+	statAppsecRule   map[string]map[string]map[string]int
+	statAlert        map[string]int
+	statStash        map[string]struct {
 		Type  string
 		Count int
 	}
@@ -62,6 +63,7 @@ func NewMetricStore() metricStore {
 		"stash":          statStash{},
 		"appsec-engine":  statAppsecEngine{},
 		"appsec-rule":    statAppsecRule{},
+		"whitelists":     statWhitelist{},
 	}
 }
 
@@ -111,6 +113,7 @@ func (ms metricStore) Fetch(url string) error {
 	mAppsecRule := ms["appsec-rule"].(statAppsecRule)
 	mAlert := ms["alerts"].(statAlert)
 	mStash := ms["stash"].(statStash)
+	mWhitelist := ms["whitelists"].(statWhitelist)
 
 	for idx, fam := range result {
 		if !strings.HasPrefix(fam.Name, "cs_") {
@@ -160,7 +163,9 @@ func (ms metricStore) Fetch(url string) error {
 
 			ival := int(fval)
 			switch fam.Name {
-			/*buckets*/
+			//
+			// buckets
+			//
 			case "cs_bucket_created_total":
 				if _, ok := mBucket[name]; !ok {
 					mBucket[name] = make(map[string]int)
@@ -190,7 +195,9 @@ func (ms metricStore) Fetch(url string) error {
 					mBucket[name] = make(map[string]int)
 				}
 				mBucket[name]["underflow"] += ival
-				/*acquis*/
+			//
+			// parsers
+			//
 			case "cs_parser_hits_total":
 				if _, ok := mAcquis[source]; !ok {
 					mAcquis[source] = make(map[string]int)
@@ -221,6 +228,33 @@ func (ms metricStore) Fetch(url string) error {
 					mParser[name] = make(map[string]int)
 				}
 				mParser[name]["unparsed"] += ival
+			//
+			// whitelists
+			//
+			case "cs_node_wl_hits_total":
+				if _, ok := mWhitelist[name]; !ok {
+					mWhitelist[name] = make(map[string]map[string]int)
+				}
+				if _, ok := mWhitelist[name][reason]; !ok {
+					mWhitelist[name][reason] = make(map[string]int)
+				}
+				mWhitelist[name][reason]["hits"] += ival
+			case "cs_node_wl_hits_ok_total":
+				if _, ok := mWhitelist[name]; !ok {
+					mWhitelist[name] = make(map[string]map[string]int)
+				}
+				if _, ok := mWhitelist[name][reason]; !ok {
+					mWhitelist[name][reason] = make(map[string]int)
+				}
+				mWhitelist[name][reason]["whitelisted"] += ival
+				// track as well whitelisted lines at acquis level
+				if _, ok := mAcquis[source]; !ok {
+					mAcquis[source] = make(map[string]int)
+				}
+				mAcquis[source]["whitelisted"] += ival
+			//
+			// lapi
+			//
 			case "cs_lapi_route_requests_total":
 				if _, ok := mLapi[route]; !ok {
 					mLapi[route] = make(map[string]int)
@@ -256,6 +290,9 @@ func (ms metricStore) Fetch(url string) error {
 					x.NonEmpty += ival
 				}
 				mLapiDecision[bouncer] = x
+			//
+			// decisions
+			//
 			case "cs_active_decisions":
 				if _, ok := mDecision[reason]; !ok {
 					mDecision[reason] = make(map[string]map[string]int)
@@ -265,15 +302,18 @@ func (ms metricStore) Fetch(url string) error {
 				}
 				mDecision[reason][origin][action] += ival
 			case "cs_alerts":
-				/*if _, ok := mAlert[scenario]; !ok {
-					mAlert[scenario] = make(map[string]int)
-				}*/
 				mAlert[reason] += ival
+			//
+			// stash
+			//
 			case "cs_cache_size":
 				mStash[name] = struct {
 					Type  string
 					Count int
 				}{Type: mtype, Count: ival}
+			//
+			// appsec
+			//
 			case "cs_appsec_reqs_total":
 				if _, ok := mAppsecEngine[metric.Labels["appsec_engine"]]; !ok {
 					mAppsecEngine[metric.Labels["appsec_engine"]] = make(map[string]int, 0)
@@ -392,15 +432,15 @@ func (cli *cliMetrics) show(sections []string, url string, noUnit bool) error {
 
 func (cli *cliMetrics) NewCommand() *cobra.Command {
 	var (
-		url string
+		url    string
 		noUnit bool
 	)
 
 	cmd := &cobra.Command{
-		Use:               "metrics",
-		Short:             "Display crowdsec prometheus metrics.",
-		Long:              `Fetch metrics from a Local API server and display them`,
-		Example:	   `# Show all Metrics, skip empty tables (same as "cecli metrics show")
+		Use:   "metrics",
+		Short: "Display crowdsec prometheus metrics.",
+		Long:  `Fetch metrics from a Local API server and display them`,
+		Example: `# Show all Metrics, skip empty tables (same as "cecli metrics show")
 cscli metrics
 
 # Show only some metrics, connect to a different url
@@ -431,7 +471,7 @@ func (cli *cliMetrics) expandSectionGroups(args []string) []string {
 	for _, section := range args {
 		switch section {
 		case "engine":
-			ret = append(ret, "acquisition", "parsers", "buckets", "stash")
+			ret = append(ret, "acquisition", "parsers", "buckets", "stash", "whitelists")
 		case "lapi":
 			ret = append(ret, "alerts", "decisions", "lapi", "lapi-bouncer", "lapi-decisions", "lapi-machine")
 		case "appsec":
@@ -446,15 +486,15 @@ func (cli *cliMetrics) expandSectionGroups(args []string) []string {
 
 func (cli *cliMetrics) newShowCmd() *cobra.Command {
 	var (
-		url string
+		url    string
 		noUnit bool
 	)
 
 	cmd := &cobra.Command{
-		Use:               "show [type]...",
-		Short:             "Display all or part of the available metrics.",
-		Long:              `Fetch metrics from a Local API server and display them, optionally filtering on specific types.`,
-		Example:	   `# Show all Metrics, skip empty tables
+		Use:   "show [type]...",
+		Short: "Display all or part of the available metrics.",
+		Long:  `Fetch metrics from a Local API server and display them, optionally filtering on specific types.`,
+		Example: `# Show all Metrics, skip empty tables
 cscli metrics show
 
 # Use an alias: "engine", "lapi" or "appsec" to show a group of metrics
@@ -482,9 +522,9 @@ cscli metrics show acquisition parsers buckets stash -o json`,
 
 func (cli *cliMetrics) list() error {
 	type metricType struct {
-		Type     string		`json:"type" yaml:"type"`
-		Title    string		`json:"title" yaml:"title"`
-		Description string	`json:"description" yaml:"description"`
+		Type        string `json:"type" yaml:"type"`
+		Title       string `json:"title" yaml:"title"`
+		Description string `json:"description" yaml:"description"`
 	}
 
 	var allMetrics []metricType

cmd/crowdsec-cli/metrics_table.go

@@ -45,6 +45,38 @@ func lapiMetricsToTable(t *table.Table, stats map[string]map[string]map[string]i
 	return numRows
 }
 
+func wlMetricsToTable(t *table.Table, stats map[string]map[string]map[string]int, noUnit bool) (int, error) {
+	if t == nil {
+		return 0, fmt.Errorf("nil table")
+	}
+
+	numRows := 0
+
+	for _, name := range maptools.SortedKeys(stats) {
+		for _, reason := range maptools.SortedKeys(stats[name]) {
+			row := make([]string, 4)
+			row[0] = name
+			row[1] = reason
+			row[2] = "-"
+			row[3] = "-"
+
+			for _, action := range maptools.SortedKeys(stats[name][reason]) {
+				value := stats[name][reason][action]
+				if action == "whitelisted" {
+					row[3] = fmt.Sprintf("%d", value)
+				} else if action == "hits" {
+					row[2] = fmt.Sprintf("%d", value)
+				} else {
+					log.Debugf("unexpected counter '%s' for whitelists = %d", action, value)
+				}
+			}
+			t.AddRow(row...)
+			numRows++
+		}
+	}
+	return numRows, nil
+}
+
 func metricsToTable(t *table.Table, stats map[string]map[string]int, keys []string, noUnit bool) (int, error) {
 	if t == nil {
 		return 0, fmt.Errorf("nil table")
@@ -95,7 +127,7 @@ func (s statBucket) Table(out io.Writer, noUnit bool, showEmpty bool) {
 		log.Warningf("while collecting bucket stats: %s", err)
 	} else if numRows > 0 || showEmpty {
 		title, _ := s.Description()
-		renderTableTitle(out, "\n" + title + ":")
+		renderTableTitle(out, "\n"+title+":")
 		t.Render()
 	}
 }
@@ -108,16 +140,16 @@ func (s statAcquis) Description() (string, string) {
 func (s statAcquis) Table(out io.Writer, noUnit bool, showEmpty bool) {
 	t := newTable(out)
 	t.SetRowLines(false)
-	t.SetHeaders("Source", "Lines read", "Lines parsed", "Lines unparsed", "Lines poured to bucket")
+	t.SetHeaders("Source", "Lines read", "Lines parsed", "Lines unparsed", "Lines poured to bucket", "Lines whitelisted")
 	t.SetAlignment(table.AlignLeft, table.AlignLeft, table.AlignLeft, table.AlignLeft, table.AlignLeft)
 
-	keys := []string{"reads", "parsed", "unparsed", "pour"}
+	keys := []string{"reads", "parsed", "unparsed", "pour", "whitelisted"}
 
 	if numRows, err := metricsToTable(t, s, keys, noUnit); err != nil {
 		log.Warningf("while collecting acquis stats: %s", err)
 	} else if numRows > 0 || showEmpty {
 		title, _ := s.Description()
-		renderTableTitle(out, "\n" + title + ":")
+		renderTableTitle(out, "\n"+title+":")
 		t.Render()
 	}
 }
@@ -137,7 +169,7 @@ func (s statAppsecEngine) Table(out io.Writer, noUnit bool, showEmpty bool) {
 		log.Warningf("while collecting appsec stats: %s", err)
 	} else if numRows > 0 || showEmpty {
 		title, _ := s.Description()
-		renderTableTitle(out, "\n" + title + ":")
+		renderTableTitle(out, "\n"+title+":")
 		t.Render()
 	}
 }
@@ -156,7 +188,7 @@ func (s statAppsecRule) Table(out io.Writer, noUnit bool, showEmpty bool) {
 		keys := []string{"triggered"}
 		if numRows, err := metricsToTable(t, appsecEngineRulesStats, keys, noUnit); err != nil {
 			log.Warningf("while collecting appsec rules stats: %s", err)
-		} else if numRows > 0 || showEmpty{
+		} else if numRows > 0 || showEmpty {
 			renderTableTitle(out, fmt.Sprintf("\nAppsec '%s' Rules Metrics:", appsecEngine))
 			t.Render()
 		}
@@ -164,6 +196,26 @@ func (s statAppsecRule) Table(out io.Writer, noUnit bool, showEmpty bool) {
 
 }
 
+func (s statWhitelist) Description() (string, string) {
+	return "Whitelist Metrics",
+		`Tracks the number of events processed and possibly whitelisted by each parser whitelist.`
+}
+
+func (s statWhitelist) Table(out io.Writer, noUnit bool, showEmpty bool) {
+	t := newTable(out)
+	t.SetRowLines(false)
+	t.SetHeaders("Whitelist", "Reason", "Hits", "Whitelisted")
+	t.SetAlignment(table.AlignLeft, table.AlignLeft, table.AlignLeft, table.AlignLeft)
+
+	if numRows, err := wlMetricsToTable(t, s, noUnit); err != nil {
+		log.Warningf("while collecting parsers stats: %s", err)
+	} else if numRows > 0 || showEmpty {
+		title, _ := s.Description()
+		renderTableTitle(out, "\n"+title+":")
+		t.Render()
+	}
+}
+
 func (s statParser) Description() (string, string) {
 	return "Parser Metrics",
 		`Tracks the number of events processed by each parser and indicates success of failure. Zero parsed lines means the parer(s) failed. Non-zero unparsed lines are fine as crowdsec select relevant lines.`
@@ -181,7 +233,7 @@ func (s statParser) Table(out io.Writer, noUnit bool, showEmpty bool) {
 		log.Warningf("while collecting parsers stats: %s", err)
 	} else if numRows > 0 || showEmpty {
 		title, _ := s.Description()
-		renderTableTitle(out, "\n" + title + ":")
+		renderTableTitle(out, "\n"+title+":")
 		t.Render()
 	}
 }
@@ -213,7 +265,7 @@ func (s statStash) Table(out io.Writer, noUnit bool, showEmpty bool) {
 	}
 	if numRows > 0 || showEmpty {
 		title, _ := s.Description()
-		renderTableTitle(out, "\n" + title + ":")
+		renderTableTitle(out, "\n"+title+":")
 		t.Render()
 	}
 }
@@ -254,7 +306,7 @@ func (s statLapi) Table(out io.Writer, noUnit bool, showEmpty bool) {
 
 	if numRows > 0 || showEmpty {
 		title, _ := s.Description()
-		renderTableTitle(out, "\n" + title + ":")
+		renderTableTitle(out, "\n"+title+":")
 		t.Render()
 	}
 }
@@ -272,9 +324,9 @@ func (s statLapiMachine) Table(out io.Writer, noUnit bool, showEmpty bool) {
 
 	numRows := lapiMetricsToTable(t, s)
 
-	if numRows > 0 || showEmpty{
+	if numRows > 0 || showEmpty {
 		title, _ := s.Description()
-		renderTableTitle(out, "\n" + title + ":")
+		renderTableTitle(out, "\n"+title+":")
 		t.Render()
 	}
 }
@@ -294,7 +346,7 @@ func (s statLapiBouncer) Table(out io.Writer, noUnit bool, showEmpty bool) {
 
 	if numRows > 0 || showEmpty {
 		title, _ := s.Description()
-		renderTableTitle(out, "\n" + title + ":")
+		renderTableTitle(out, "\n"+title+":")
 		t.Render()
 	}
 }
@@ -320,9 +372,9 @@ func (s statLapiDecision) Table(out io.Writer, noUnit bool, showEmpty bool) {
 		numRows++
 	}
 
-	if numRows > 0 || showEmpty{
+	if numRows > 0 || showEmpty {
 		title, _ := s.Description()
-		renderTableTitle(out, "\n" + title + ":")
+		renderTableTitle(out, "\n"+title+":")
 		t.Render()
 	}
 }
@@ -353,9 +405,9 @@ func (s statDecision) Table(out io.Writer, noUnit bool, showEmpty bool) {
 		}
 	}
 
-	if numRows > 0 || showEmpty{
+	if numRows > 0 || showEmpty {
 		title, _ := s.Description()
-		renderTableTitle(out, "\n" + title + ":")
+		renderTableTitle(out, "\n"+title+":")
 		t.Render()
 	}
 }
@@ -380,9 +432,9 @@ func (s statAlert) Table(out io.Writer, noUnit bool, showEmpty bool) {
 		numRows++
 	}
 
-	if numRows > 0 || showEmpty{
+	if numRows > 0 || showEmpty {
 		title, _ := s.Description()
-		renderTableTitle(out, "\n" + title + ":")
+		renderTableTitle(out, "\n"+title+":")
 		t.Render()
 	}
 }

cmd/crowdsec/metrics.go

@@ -161,7 +161,7 @@ func registerPrometheus(config *csconfig.PrometheusCfg) {
 			leaky.BucketsUnderflow, leaky.BucketsCanceled, leaky.BucketsInstantiation, leaky.BucketsOverflow,
 			v1.LapiRouteHits,
 			leaky.BucketsCurrentCount,
-			cache.CacheMetrics, exprhelpers.RegexpCacheMetrics,
+			cache.CacheMetrics, exprhelpers.RegexpCacheMetrics, parser.NodesWlHitsOk, parser.NodesWlHits,
 		)
 	} else {
 		log.Infof("Loading prometheus collectors")
@@ -170,7 +170,7 @@ func registerPrometheus(config *csconfig.PrometheusCfg) {
 			globalCsInfo, globalParsingHistogram, globalPourHistogram,
 			v1.LapiRouteHits, v1.LapiMachineHits, v1.LapiBouncerHits, v1.LapiNilDecisions, v1.LapiNonNilDecisions, v1.LapiResponseTime,
 			leaky.BucketsPour, leaky.BucketsUnderflow, leaky.BucketsCanceled, leaky.BucketsInstantiation, leaky.BucketsOverflow, leaky.BucketsCurrentCount,
-			globalActiveDecisions, globalAlerts,
+			globalActiveDecisions, globalAlerts, parser.NodesWlHitsOk, parser.NodesWlHits,
 			cache.CacheMetrics, exprhelpers.RegexpCacheMetrics,
 		)
 

pkg/parser/node.go

@@ -168,9 +168,9 @@ func (n *Node) process(p *types.Event, ctx UnixParserCtx, expressionEnv map[stri
 		NodesHits.With(prometheus.Labels{"source": p.Line.Src, "type": p.Line.Module, "name": n.Name}).Inc()
 	}
 	exprErr := error(nil)
-	isWhitelisted := n.CheckIPsWL(p.ParseIPSources())
+	isWhitelisted := n.CheckIPsWL(p)
 	if !isWhitelisted {
-		isWhitelisted, exprErr = n.CheckExprWL(cachedExprEnv)
+		isWhitelisted, exprErr = n.CheckExprWL(cachedExprEnv, p)
 	}
 	if exprErr != nil {
 		// Previous code returned nil if there was an error, so we keep this behavior

pkg/parser/runtime.go

@@ -221,6 +221,24 @@ var NodesHitsKo = prometheus.NewCounterVec(
 	[]string{"source", "type", "name"},
 )
 
+//
+
+var NodesWlHitsOk = prometheus.NewCounterVec(
+	prometheus.CounterOpts{
+		Name: "cs_node_wl_hits_ok_total",
+		Help: "Total events successfully whitelisted by node.",
+	},
+	[]string{"source", "type", "name", "reason"},
+)
+
+var NodesWlHits = prometheus.NewCounterVec(
+	prometheus.CounterOpts{
+		Name: "cs_node_wl_hits_total",
+		Help: "Total events processed by whitelist node.",
+	},
+	[]string{"source", "type", "name", "reason"},
+)
+
 func stageidx(stage string, stages []string) int {
 	for i, v := range stages {
 		if stage == v {

pkg/parser/whitelist.go

@@ -8,6 +8,7 @@ import (
 	"github.com/antonmedv/expr/vm"
 	"github.com/crowdsecurity/crowdsec/pkg/exprhelpers"
 	"github.com/crowdsecurity/crowdsec/pkg/types"
+	"github.com/prometheus/client_golang/prometheus"
 )
 
 type Whitelist struct {
@@ -36,11 +37,13 @@ func (n *Node) ContainsIPLists() bool {
 	return len(n.Whitelist.B_Ips) > 0 || len(n.Whitelist.B_Cidrs) > 0
 }
 
-func (n *Node) CheckIPsWL(srcs []net.IP) bool {
+func (n *Node) CheckIPsWL(p *types.Event) bool {
+	srcs := p.ParseIPSources()
 	isWhitelisted := false
 	if !n.ContainsIPLists() {
 		return isWhitelisted
 	}
+	NodesWlHits.With(prometheus.Labels{"source": p.Line.Src, "type": p.Line.Module, "name": n.Name, "reason": n.Whitelist.Reason}).Inc()
 	for _, src := range srcs {
 		if isWhitelisted {
 			break
@@ -62,15 +65,19 @@ func (n *Node) CheckIPsWL(srcs []net.IP) bool {
 			n.Logger.Tracef("whitelist: %s not in [%s]", src, v)
 		}
 	}
+	if isWhitelisted {
+		NodesWlHitsOk.With(prometheus.Labels{"source": p.Line.Src, "type": p.Line.Module, "name": n.Name, "reason": n.Whitelist.Reason}).Inc()
+	}
 	return isWhitelisted
 }
 
-func (n *Node) CheckExprWL(cachedExprEnv map[string]interface{}) (bool, error) {
+func (n *Node) CheckExprWL(cachedExprEnv map[string]interface{}, p *types.Event) (bool, error) {
 	isWhitelisted := false
 
 	if !n.ContainsExprLists() {
 		return false, nil
 	}
+	NodesWlHits.With(prometheus.Labels{"source": p.Line.Src, "type": p.Line.Module, "name": n.Name, "reason": n.Whitelist.Reason}).Inc()
 	/* run whitelist expression tests anyway */
 	for eidx, e := range n.Whitelist.B_Exprs {
 		//if we already know the event is whitelisted, skip the rest of the expressions
@@ -94,6 +101,9 @@ func (n *Node) CheckExprWL(cachedExprEnv map[string]interface{}) (bool, error) {
 			n.Logger.Errorf("unexpected type %t (%v) while running '%s'", output, output, n.Whitelist.Exprs[eidx])
 		}
 	}
+	if isWhitelisted {
+		NodesWlHitsOk.With(prometheus.Labels{"source": p.Line.Src, "type": p.Line.Module, "name": n.Name, "reason": n.Whitelist.Reason}).Inc()
+	}
 	return isWhitelisted, nil
 }
 

pkg/parser/whitelist_test.go

@@ -289,9 +289,9 @@ func TestWhitelistCheck(t *testing.T) {
 			var err error
 			node.Whitelist = tt.whitelist
 			node.CompileWLs()
-			isWhitelisted := node.CheckIPsWL(tt.event.ParseIPSources())
+			isWhitelisted := node.CheckIPsWL(tt.event)
 			if !isWhitelisted {
-				isWhitelisted, err = node.CheckExprWL(map[string]interface{}{"evt": tt.event})
+				isWhitelisted, err = node.CheckExprWL(map[string]interface{}{"evt": tt.event}, tt.event)
 			}
 			require.NoError(t, err)
 			require.Equal(t, tt.expected, isWhitelisted)