Add a new input plugin for InfiniBand card/port statistics

Gopkg.toml

@@ -297,6 +297,10 @@
   branch = "master"
   name = "github.com/cisco-ie/nx-telemetry-proto"
 
+[[constraint]]
+  branch = "master"
+  name = "github.com/Mellanox/rdmamap"
+
 [[constraint]]
   name = "gopkg.in/ldap.v3"
   version = "3.1.0"

plugins/inputs/all/all.go

@@ -56,6 +56,7 @@ import (
 	_ "github.com/influxdata/telegraf/plugins/inputs/http_response"
 	_ "github.com/influxdata/telegraf/plugins/inputs/httpjson"
 	_ "github.com/influxdata/telegraf/plugins/inputs/icinga2"
+	_ "github.com/influxdata/telegraf/plugins/inputs/infiniband"
 	_ "github.com/influxdata/telegraf/plugins/inputs/influxdb"
 	_ "github.com/influxdata/telegraf/plugins/inputs/influxdb_listener"
 	_ "github.com/influxdata/telegraf/plugins/inputs/internal"

plugins/inputs/infiniband/README.md

@@ -0,0 +1,29 @@
+# InfiniBand Input Plugin
+
+This plugin gathers statistics for all InfiniBand devices and ports on the system. These are the counters that can be found in /sys/class/infiniband/<dev>/port/<port>/counters/
+
+### Configuration
+
+This section contains the default TOML to configure the plugin.  You can
+generate it using `telegraf --usage infiniband`.
+
+```toml
+[[inputs.infiniband]]
+```
+
+There are no configuration options for this plugin.
+
+### Metrics
+
+You can find more information about the counters that are gathered here: 
+https://community.mellanox.com/s/article/understanding-mlx5-linux-counters-and-status-parameters
+
+There is a simple mapping from counter -> counter value. All counter values are 64 bit integers. A seperate measurement is made for each port.
+Each measurement is tagged with the device and port that it relates to. These are strings.
+
+
+### Example Output
+
+```
+infiniband,device=mlx5_0,port=1,VL15_dropped=0i,excessive_buffer_overrun_errors=0i,link_downed=0i,link_error_recovery=0i,local_link_integrity_errors=0i,multicast_rcv_packets=0i,multicast_xmit_packets=0i,port_rcv_constraint_errors=0i,port_rcv_data=237159415345822i,port_rcv_errors=0i,port_rcv_packets=801977655075i,port_rcv_remote_physical_errors=0i,port_rcv_switch_relay_errors=0i,port_xmit_constraint_errors=0i,port_xmit_data=238334949937759i,port_xmit_discards=0i,port_xmit_packets=803162651391i,port_xmit_wait=4294967295i,symbol_error=0i,unicast_rcv_packets=801977655075i,unicast_xmit_packets=803162651391i 1573125558000000000
+```

plugins/inputs/infiniband/infiniband.go

@@ -0,0 +1,22 @@
+package infiniband
+
+import (
+	"github.com/influxdata/telegraf"
+)
+
+// Stores the configuration values for the infiniband plugin - as there are no
+// config values, this is intentionally empty
+type Infiniband struct {
+	Log telegraf.Logger `toml:"-"`
+}
+
+// Sample configuration for plugin
+var InfinibandConfig = ``
+
+func (_ *Infiniband) SampleConfig() string {
+	return InfinibandConfig
+}
+
+func (_ *Infiniband) Description() string {
+	return "Gets counters from all InfiniBand cards and ports installed"
+}

plugins/inputs/infiniband/infiniband_linux.go

@@ -0,0 +1,59 @@
+// +build linux
+
+package infiniband
+
+import (
+	"fmt"
+	"github.com/Mellanox/rdmamap"
+	"github.com/influxdata/telegraf"
+	"github.com/influxdata/telegraf/plugins/inputs"
+	"strconv"
+)
+
+// Gather statistics from our infiniband cards
+func (_ *Infiniband) Gather(acc telegraf.Accumulator) error {
+
+	rdmaDevices := rdmamap.GetRdmaDeviceList()
+
+	if len(rdmaDevices) == 0 {
+		return fmt.Errorf("no InfiniBand devices found in /sys/class/infiniband/")
+	}
+
+	for _, dev := range rdmaDevices {
+		devicePorts := rdmamap.GetPorts(dev)
+		for _, port := range devicePorts {
+			portInt, err := strconv.Atoi(port)
+			if err != nil {
+				return err
+			}
+
+			stats, err := rdmamap.GetRdmaSysfsStats(dev, portInt)
+			if err != nil {
+				return err
+			}
+
+			addStats(dev, port, stats, acc)
+		}
+	}
+
+	return nil
+}
+
+// Add the statistics to the accumulator
+func addStats(dev string, port string, stats []rdmamap.RdmaStatEntry, acc telegraf.Accumulator) {
+
+	// Allow users to filter by card and port
+	tags := map[string]string{"device": dev, "port": port}
+	fields := make(map[string]interface{})
+
+	for _, entry := range stats {
+		fields[entry.Name] = entry.Value
+	}
+
+	acc.AddFields("infiniband", fields, tags)
+}
+
+// Initialise plugin
+func init() {
+	inputs.Add("infiniband", func() telegraf.Input { return &Infiniband{} })
+}

plugins/inputs/infiniband/infiniband_notlinux.go

@@ -0,0 +1,23 @@
+// +build !linux
+
+package infiniband
+
+import (
+	"github.com/influxdata/telegraf"
+	"github.com/influxdata/telegraf/plugins/inputs"
+)
+
+func (i *Infiniband) Init() error {
+	i.Log.Warn("Current platform is not supported")
+	return nil
+}
+
+func (_ *Infiniband) Gather(acc telegraf.Accumulator) error {
+	return nil
+}
+
+func init() {
+	inputs.Add("infiniband", func() telegraf.Input {
+		return &Infiniband{}
+	})
+}

plugins/inputs/infiniband/infiniband_test.go

@@ -0,0 +1,134 @@
+// +build linux
+
+package infiniband
+
+import (
+	"github.com/Mellanox/rdmamap"
+	"github.com/influxdata/telegraf/testutil"
+	"testing"
+)
+
+func TestInfiniband(t *testing.T) {
+	fields := map[string]interface{}{
+		"excessive_buffer_overrun_errors": uint64(0),
+		"link_downed":                     uint64(0),
+		"link_error_recovery":             uint64(0),
+		"local_link_integrity_errors":     uint64(0),
+		"multicast_rcv_packets":           uint64(0),
+		"multicast_xmit_packets":          uint64(0),
+		"port_rcv_constraint_errors":      uint64(0),
+		"port_rcv_data":                   uint64(237159415345822),
+		"port_rcv_errors":                 uint64(0),
+		"port_rcv_packets":                uint64(801977655075),
+		"port_rcv_remote_physical_errors": uint64(0),
+		"port_rcv_switch_relay_errors":    uint64(0),
+		"port_xmit_constraint_errors":     uint64(0),
+		"port_xmit_data":                  uint64(238334949937759),
+		"port_xmit_discards":              uint64(0),
+		"port_xmit_packets":               uint64(803162651391),
+		"port_xmit_wait":                  uint64(4294967295),
+		"symbol_error":                    uint64(0),
+		"unicast_rcv_packets":             uint64(801977655075),
+		"unicast_xmit_packets":            uint64(803162651391),
+		"VL15_dropped":                    uint64(0),
+	}
+
+	tags := map[string]string{
+		"device": "m1x5_0",
+		"port":   "1",
+	}
+
+	sample_rdmastats_entries := []rdmamap.RdmaStatEntry{
+		{
+			Name:  "excessive_buffer_overrun_errors",
+			Value: uint64(0),
+		},
+		{
+			Name:  "link_downed",
+			Value: uint64(0),
+		},
+		{
+			Name:  "link_error_recovery",
+			Value: uint64(0),
+		},
+		{
+			Name:  "local_link_integrity_errors",
+			Value: uint64(0),
+		},
+		{
+			Name:  "multicast_rcv_packets",
+			Value: uint64(0),
+		},
+		{
+			Name:  "multicast_xmit_packets",
+			Value: uint64(0),
+		},
+		{
+			Name:  "port_rcv_constraint_errors",
+			Value: uint64(0),
+		},
+		{
+			Name:  "port_rcv_data",
+			Value: uint64(237159415345822),
+		},
+		{
+			Name:  "port_rcv_errors",
+			Value: uint64(0),
+		},
+		{
+			Name:  "port_rcv_packets",
+			Value: uint64(801977655075),
+		},
+		{
+			Name:  "port_rcv_remote_physical_errors",
+			Value: uint64(0),
+		},
+		{
+			Name:  "port_rcv_switch_relay_errors",
+			Value: uint64(0),
+		},
+		{
+			Name:  "port_xmit_constraint_errors",
+			Value: uint64(0),
+		},
+		{
+			Name:  "port_xmit_data",
+			Value: uint64(238334949937759),
+		},
+		{
+			Name:  "port_xmit_discards",
+			Value: uint64(0),
+		},
+		{
+			Name:  "port_xmit_packets",
+			Value: uint64(803162651391),
+		},
+		{
+			Name:  "port_xmit_wait",
+			Value: uint64(4294967295),
+		},
+		{
+			Name:  "symbol_error",
+			Value: uint64(0),
+		},
+		{
+			Name:  "unicast_rcv_packets",
+			Value: uint64(801977655075),
+		},
+		{
+			Name:  "unicast_xmit_packets",
+			Value: uint64(803162651391),
+		},
+		{
+			Name:  "VL15_dropped",
+			Value: uint64(0),
+		},
+	}
+
+	var acc testutil.Accumulator
+
+	addStats("m1x5_0", "1", sample_rdmastats_entries, &acc)
+
+	acc.AssertContainsTaggedFields(t, "infiniband", fields, tags)
+
+}