Merge pull request alibaba#160 from Lyt99/fix/probes-dashboard

format metric names & dashboard use new metrics

deploy/skoopbundle.yaml

@@ -131,7 +131,6 @@ data:
       - name: tcp
       - name: tcpext
       - name: udp
-      - name: packetloss
       - name: flow
     event:
       probes:

pkg/exporter/cmd/eventserver.go

@@ -110,6 +110,7 @@ func (m *EventProbeManager) CreateProbe(config ProbeConfig) (probe.EventProbe, e
 }
 
 func (m *EventProbeManager) StartProbe(ctx context.Context, probe probe.EventProbe) error {
+	log.Infof("start event probe %s", probe.Name())
 	return probe.Start(ctx)
 }
 

pkg/exporter/probe/procio/procio.go

@@ -14,10 +14,10 @@ import (
 )
 
 const (
-	IOReadSyscall  = "ioreadsyscall"
-	IOWriteSyscall = "iowritesyscall"
-	IOReadBytes    = "ioreadbytes"
-	IOWriteBytes   = "iowritebytes"
+	IOReadSyscall  = "readsyscall"
+	IOWriteSyscall = "writesyscall"
+	IOReadBytes    = "readbytes"
+	IOWriteBytes   = "writebytes"
 
 	probeName = "io" // nolint
 )

pkg/exporter/probe/tracekernel/tracekernel.go

@@ -40,10 +40,10 @@ const (
 	HOOK_IPOUTPUT    = "ip_output"
 	HOOK_IPOUTPUTFIN = "ip_finish_output2"
 
-	RXKERNEL_SLOW_METRIC      = "kernellatency_rxslow"
-	TXKERNEL_SLOW_METRIC      = "kernellatency_txslow"
-	RXKERNEL_SLOW100MS_METRIC = "kernellatency_rxslow100ms"
-	TXKERNEL_SLOW100MS_METRIC = "kernellatency_txslow100ms"
+	RXKERNEL_SLOW_METRIC      = "rxslow"
+	TXKERNEL_SLOW_METRIC      = "txslow"
+	RXKERNEL_SLOW100MS_METRIC = "rxslow100ms"
+	TXKERNEL_SLOW100MS_METRIC = "txslow100ms"
 
 	probeTypeEvent   = 0
 	probeTypeMetrics = 1

pkg/exporter/probe/tracenetiftxlatency/tracenetiftxlatency.go

@@ -27,8 +27,8 @@ import (
 //
 //go:generate go run github.com/cilium/ebpf/cmd/bpf2go -cc clang -cflags $BPF_CFLAGS -type insp_nftxlat_event_t -type insp_nftxlat_metric_t bpf ../../../../bpf/netiftxlatency.c -- -I../../../../bpf/headers -D__TARGET_ARCH_x86
 const (
-	TXLAT_QDISC_SLOW  = "netiftxlat_qdiscslow100ms"
-	TXLAT_NETDEV_SLOW = "netiftxlat_netdevslow100ms"
+	TXLAT_QDISC_SLOW  = "qdiscslow100ms"
+	TXLAT_NETDEV_SLOW = "netdevslow100ms"
 )
 
 var (

pkg/exporter/probe/tracenetsoftirq/tracenetsoftirq.go

@@ -24,16 +24,18 @@ import (
 
 //go:generate go run github.com/cilium/ebpf/cmd/bpf2go -cc clang -cflags $BPF_CFLAGS -type insp_softirq_event_t bpf ../../../../bpf/net_softirq.c -- -I../../../../bpf/headers -D__TARGET_ARCH_x86
 const (
-	NETSOFTIRQ_SCHED_SLOW   = "net_softirq_schedslow"       //nolint
-	NETSOFTIRQ_SCHED_100MS  = "net_softirq_schedslow100ms"  //nolint
-	NETSOFTIRQ_EXCUTE_SLOW  = "net_softirq_excuteslow"      //nolint
-	NETSOFTIRQ_EXCUTE_100MS = "net_softirq_excuteslow100ms" //nolint
+	NETSOFTIRQ_SCHED_SLOW   = "schedslow"       //nolint
+	NETSOFTIRQ_SCHED_100MS  = "schedslow100ms"  //nolint
+	NETSOFTIRQ_EXCUTE_SLOW  = "excuteslow"      //nolint
+	NETSOFTIRQ_EXCUTE_100MS = "excuteslow100ms" //nolint
 )
 
 var (
 	metrics          = []string{NETSOFTIRQ_SCHED_SLOW, NETSOFTIRQ_SCHED_100MS, NETSOFTIRQ_EXCUTE_SLOW, NETSOFTIRQ_EXCUTE_100MS}
 	probeName        = "netsoftirq"
-	_netSoftirqProbe = &netSoftirqProbe{}
+	_netSoftirqProbe = &netSoftirqProbe{
+		metricsMap: make(map[string]map[uint32]uint64),
+	}
 )
 
 func init() {

pkg/exporter/probe/tracepacketloss/packetloss.go

@@ -173,17 +173,17 @@ func (p *packetLossProbe) start(ctx context.Context, probeType probe.Type) (err
 			_ = p.cleanup()
 			return
 		}
+	}
 
-		if p.refcnt[probe.ProbeTypeEvent] == 1 {
-			p.perfReader, err = perf.NewReader(p.objs.bpfMaps.InspPlEvent, int(unsafe.Sizeof(bpfInspPlEventT{})))
-			if err != nil {
-				log.Errorf("%s error create perf reader, err: %v", probeName, err)
-				_ = p.stop(ctx, probeType)
-				return
-			}
-
-			go p.perfLoop()
+	if p.refcnt[probe.ProbeTypeEvent] == 1 {
+		p.perfReader, err = perf.NewReader(p.objs.bpfMaps.InspPlEvent, int(unsafe.Sizeof(bpfInspPlEventT{})))
+		if err != nil {
+			log.Errorf("%s error create perf reader, err: %v", probeName, err)
+			_ = p.stop(ctx, probeType)
+			return
 		}
+
+		go p.perfLoop()
 	}
 	return nil
 }
@@ -343,9 +343,9 @@ func (p *packetLossProbe) perfLoop() {
 			strs = append(strs, sym.GetExpr())
 		}
 
-		stackStr := strings.Join(strs, " ")
+		stackStr := strings.Join(strs, "\n")
 
-		evt.Message = fmt.Sprintf("%s stacktrace:%s", tuple, stackStr)
+		evt.Message = fmt.Sprintf("%s\nstacktrace:\n%s", tuple, stackStr)
 
 		if p.sink != nil {
 			log.Debugf("%s sink event %s", probeName, util.ToJSONString(evt))

pkg/exporter/probe/tracesocketlatency/socketlatency.go

@@ -32,11 +32,11 @@ const (
 	SOCKETLAT_READSLOW = "SOCKETLAT_READSLOW"
 	SOCKETLAT_SENDSLOW = "SOCKETLAT_SENDSLOW"
 
-	READ100MS  = "socketlatencyread100ms"
-	READ300MS  = "socketlatencyread300ms"
-	READ1MS    = "socketlatencyread1ms"
-	WRITE100MS = "socketlatencywrite100ms"
-	WRITE1MS   = "socketlatencywrite1ms"
+	READ100MS  = "read100ms"
+	READ300MS  = "read300ms"
+	READ1MS    = "read1ms"
+	WRITE100MS = "write100ms"
+	WRITE1MS   = "write1ms"
 
 	/*
 		#define ACTION_READ	    1

pkg/exporter/probe/tracevirtcmdlat/tracevirtcmdlat.go

@@ -24,17 +24,19 @@ import (
 //go:generate go run github.com/cilium/ebpf/cmd/bpf2go -cc clang -cflags $BPF_CFLAGS -type insp_virtcmdlat_event_t  bpf ../../../../bpf/virtcmdlatency.c -- -I../../../../bpf/headers -D__TARGET_ARCH_x86
 
 const (
-	VIRTCMD100MS  = "virtcmdlatency100ms"
-	VIRTCMD       = "virtcmdlatency"
+	VIRTCMD100MS  = "latency100ms"
+	VIRTCMD       = "latency"
 	VIRTCMDEXCUTE = "VIRTCMDEXCUTE"
 
 	fn        = "virtnet_send_command"
-	probeName = "virtcmdLatency"
+	probeName = "virtcmdlatency"
 )
 
 var (
 	metrics              = []string{VIRTCMD100MS, VIRTCMD}
-	_virtcmdLatencyProbe = &virtcmdLatencyProbe{}
+	_virtcmdLatencyProbe = &virtcmdLatencyProbe{
+		metricsMap: make(map[string]map[uint32]uint64),
+	}
 )
 
 func init() {