CL-Softwares
/
telegraf
8
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

feat(inputs.ipmi_sensor): Collect additional commands (#15495)

This commit is contained in:
Joshua Powers 2024-06-18 14:44:43 -06:00 committed by GitHub
parent 784ede96f8
commit 6fb42764e1
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
4 changed files with 301 additions and 65 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

59
plugins/inputs/ipmi_sensor/README.md
View File

@ -44,42 +44,48 @@ See the [CONFIGURATION.md][CONFIGURATION.md] for more details.
```toml @sample.conf
# Read metrics from the bare metal servers via IPMI
[[inputs.ipmi_sensor]]
## optionally specify the path to the ipmitool executable
## Specify the path to the ipmitool executable
# path = "/usr/bin/ipmitool"
##
## Use sudo
## Setting 'use_sudo' to true will make use of sudo to run ipmitool.
## Sudo must be configured to allow the telegraf user to run ipmitool
## without a password.
# use_sudo = false
##
## optionally force session privilege level. Can be CALLBACK, USER, OPERATOR, ADMINISTRATOR
# privilege = "ADMINISTRATOR"
##
## optionally specify one or more servers via a url matching
## Servers
## Specify one or more servers via a url. If no servers are specified, local
## machine sensor stats will be queried. Uses the format:
## [username[:password]@][protocol[(address)]]
## e.g.
## root:passwd@lan(127.0.0.1)
##
## if no servers are specified, local machine sensor stats will be queried
##
## e.g. root:passwd@lan(127.0.0.1)
# servers = ["USERID:PASSW0RD@lan(192.168.1.1)"]
## Recommended: use metric 'interval' that is a multiple of 'timeout' to avoid
## gaps or overlap in pulled data
interval = "30s"
## Session privilege level
## Choose from: CALLBACK, USER, OPERATOR, ADMINISTRATOR
# privilege = "ADMINISTRATOR"
## Timeout for the ipmitool command to complete. Default is 20 seconds.
timeout = "20s"
## Timeout
## Timeout for the ipmitool command to complete.
# timeout = "20s"
## Schema Version: (Optional, defaults to version 1)
metric_version = 2
## Metric schema version
## See the plugin readme for more information on schema versioning.
# metric_version = 1
## Sensors to collect
## Choose from:
## * sdr: default, collects sensor data records
## * chassis_power_status: collects the power status of the chassis
## * dcmi_power_reading: collects the power readings from the Data Center Management Interface
# sensors = ["sdr"]
## Hex key
## Optionally provide the hex key for the IMPI connection.
# hex_key = ""
## Cache
## If ipmitool should use a cache
## for me ipmitool runs about 2 to 10 times faster with cache enabled on HP G10 servers (when using ubuntu20.04)
## the cache file may not work well for you if some sensors come up late
## Using a cache can speed up collection times depending on your device.
# use_cache = false
## Path to the ipmitools cache file (defaults to OS temp dir)
@ -87,6 +93,17 @@ See the [CONFIGURATION.md][CONFIGURATION.md] for more details.
# cache_path = ""
```
## Sensors
By default the plugin collects data via the `sdr` command and returns those
values. However, there are additonal sensor options that be call on:
- `chassis_power_status` - returns 0 or 1 depending on the output of
`chassis power status`
- `dcmi_power_reading` - Returns the watt values from `dcmi power reading`
These sensor options are not affected by the metric version.
## Metrics
Version 1 schema:

137
plugins/inputs/ipmi_sensor/ipmi_sensor.go
View File

@ -19,6 +19,7 @@ import (
"github.com/influxdata/telegraf"
"github.com/influxdata/telegraf/config"
"github.com/influxdata/telegraf/internal"
"github.com/influxdata/telegraf/internal/choice"
"github.com/influxdata/telegraf/plugins/inputs"
)
@ -31,21 +32,22 @@ var (
reV2ParseLine = regexp.MustCompile(`^(?P<name>[^|]*)\|[^|]+\|(?P<status_code>[^|]*)\|(?P<entity_id>[^|]*)\|(?:(?P<description>[^|]+))?`)
reV2ParseDescription = regexp.MustCompile(`^(?P<analogValue>-?[0-9.]+)\s(?P<analogUnit>.*)|(?P<status>.+)|^$`)
reV2ParseUnit = regexp.MustCompile(`^(?P<realAnalogUnit>[^,]+)(?:,\s*(?P<statusDesc>.*))?`)
dcmiPowerReading = regexp.MustCompile(`^(?P<name>[^|]*)\:(?P<value>.* Watts)?`)
)
// Ipmi stores the configuration values for the ipmi_sensor input plugin
type Ipmi struct {
Path string
Privilege string
HexKey string `toml:"hex_key"`
Servers []string
Timeout config.Duration
MetricVersion int
UseSudo bool
UseCache bool
CachePath string
Log telegraf.Logger `toml:"-"`
Path string `toml:"path"`
Privilege string `toml:"privilege"`
HexKey string `toml:"hex_key"`
Servers []string `toml:"servers"`
Sensors []string `toml:"sensors"`
Timeout config.Duration `toml:"timeout"`
MetricVersion int `toml:"metric_version"`
UseSudo bool `toml:"use_sudo"`
UseCache bool `toml:"use_cache"`
CachePath string `toml:"cache_path"`
Log telegraf.Logger `toml:"-"`
}
const cmd = "ipmitool"
@ -66,6 +68,12 @@ func (m *Ipmi) Init() error {
if m.CachePath == "" {
m.CachePath = os.TempDir()
}
if len(m.Sensors) == 0 {
m.Sensors = []string{"sdr"}
}
if err := choice.CheckSlice(m.Sensors, []string{"sdr", "chassis_power_status", "dcmi_power_reading"}); err != nil {
return err
}
// Check parameters
if m.Path == "" {
@ -87,24 +95,37 @@ func (m *Ipmi) Gather(acc telegraf.Accumulator) error {
wg.Add(1)
go func(a telegraf.Accumulator, s string) {
defer wg.Done()
err := m.parse(a, s)
if err != nil {
a.AddError(err)
for _, sensor := range m.Sensors {
a.AddError(m.parse(a, s, sensor))
}
}(acc, server)
}
wg.Wait()
} else {
err := m.parse(acc, "")
if err != nil {
return err
for _, sensor := range m.Sensors {
err := m.parse(acc, "", sensor)
if err != nil {
return err
}
}
}
return nil
}
func (m *Ipmi) parse(acc telegraf.Accumulator, server string) error {
func (m *Ipmi) parse(acc telegraf.Accumulator, server string, sensor string) error {
var command []string
switch sensor {
case "sdr":
command = append(command, "sdr")
case "chassis_power_status":
command = append(command, "chassis", "power", "status")
case "dcmi_power_reading":
command = append(command, "dcmi", "power", "reading")
default:
return fmt.Errorf("unknown sensor type %q", sensor)
}
opts := make([]string, 0)
hostname := ""
if server != "" {
@ -112,7 +133,9 @@ func (m *Ipmi) parse(acc telegraf.Accumulator, server string) error {
hostname = conn.Hostname
opts = conn.options()
}
opts = append(opts, "sdr")
opts = append(opts, command...)
if m.UseCache {
cacheFile := filepath.Join(m.CachePath, server+"_ipmi_cache")
_, err := os.Stat(cacheFile)
@ -134,7 +157,7 @@ func (m *Ipmi) parse(acc telegraf.Accumulator, server string) error {
}
opts = append(opts, "-S", cacheFile)
}
if m.MetricVersion == 2 {
if m.MetricVersion == 2 && sensor == "sdr" {
opts = append(opts, "elist")
}
name := m.Path
@ -149,10 +172,78 @@ func (m *Ipmi) parse(acc telegraf.Accumulator, server string) error {
if err != nil {
return fmt.Errorf("failed to run command %q: %w - %s", strings.Join(sanitizeIPMICmd(cmd.Args), " "), err, string(out))
}
if m.MetricVersion == 2 {
return m.parseV2(acc, hostname, out, timestamp)
switch sensor {
case "sdr":
if m.MetricVersion == 2 {
return m.parseV2(acc, hostname, out, timestamp)
} else {
return m.parseV1(acc, hostname, out, timestamp)
}
case "chassis_power_status":
return m.parseChassisPowerStatus(acc, hostname, out, timestamp)
case "dcmi_power_reading":
return m.parseDCMIPowerReading(acc, hostname, out, timestamp)
}
return m.parseV1(acc, hostname, out, timestamp)
return fmt.Errorf("unknown sensor type %q", sensor)
}
func (m *Ipmi) parseChassisPowerStatus(acc telegraf.Accumulator, hostname string, cmdOut []byte, measuredAt time.Time) error {
// each line will look something like
// Chassis Power is on
// Chassis Power is off
scanner := bufio.NewScanner(bytes.NewReader(cmdOut))
for scanner.Scan() {
line := scanner.Text()
if strings.Contains(line, "Chassis Power is on") {
acc.AddFields("ipmi_sensor", map[string]interface{}{"value": 1}, map[string]string{"name": "chassis_power_status", "server": hostname}, measuredAt)
} else if strings.Contains(line, "Chassis Power is off") {
acc.AddFields("ipmi_sensor", map[string]interface{}{"value": 0}, map[string]string{"name": "chassis_power_status", "server": hostname}, measuredAt)
}
}
return scanner.Err()
}
func (m *Ipmi) parseDCMIPowerReading(acc telegraf.Accumulator, hostname string, cmdOut []byte, measuredAt time.Time) error {
// each line will look something like
// Current Power Reading : 0.000
scanner := bufio.NewScanner(bytes.NewReader(cmdOut))
for scanner.Scan() {
ipmiFields := m.extractFieldsFromRegex(dcmiPowerReading, scanner.Text())
if len(ipmiFields) != 2 {
continue
}
tags := map[string]string{
"name": transform(ipmiFields["name"]),
}
// tag the server is we have one
if hostname != "" {
tags["server"] = hostname
}
fields := make(map[string]interface{})
valunit := strings.Split(ipmiFields["value"], " ")
if len(valunit) != 2 {
continue
}
var err error
fields["value"], err = aToFloat(valunit[0])
if err != nil {
continue
}
if len(valunit) > 1 {
tags["unit"] = transform(valunit[1])
}
acc.AddFields("ipmi_sensor", fields, tags, measuredAt)
}
return scanner.Err()
}
func (m *Ipmi) parseV1(acc telegraf.Accumulator, hostname string, cmdOut []byte, measuredAt time.Time) error {

122
plugins/inputs/ipmi_sensor/ipmi_sensor_test.go
View File

@ -771,6 +771,128 @@ func Test_parseV2(t *testing.T) {
}
}
func Test_parsePowerStatus(t *testing.T) {
type args struct {
hostname string
cmdOut []byte
measuredAt time.Time
}
tests := []struct {
name string
args args
expected []telegraf.Metric
}{
{
name: "Test correct parse power status off",
args: args{
hostname: "host",
cmdOut: []byte("Chassis Power is off"),
measuredAt: time.Now(),
},
expected: []telegraf.Metric{
testutil.MustMetric("ipmi_sensor",
map[string]string{
"name": "chassis_power_status",
"server": "host",
},
map[string]interface{}{"value": 0},
time.Unix(0, 0),
),
},
},
{
name: "Test correct parse power status on",
args: args{
hostname: "host",
cmdOut: []byte("Chassis Power is on"),
measuredAt: time.Now(),
},
expected: []telegraf.Metric{
testutil.MustMetric("ipmi_sensor",
map[string]string{
"name": "chassis_power_status",
"server": "host",
},
map[string]interface{}{"value": 1},
time.Unix(0, 0),
),
},
},
}
ipmi := &Ipmi{
Log: testutil.Logger{},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
var acc testutil.Accumulator
err := ipmi.parseChassisPowerStatus(&acc, tt.args.hostname, tt.args.cmdOut, tt.args.measuredAt)
require.NoError(t, err)
testutil.RequireMetricsEqual(t, tt.expected, acc.GetTelegrafMetrics(), testutil.IgnoreTime())
})
}
}
func Test_parsePowerReading(t *testing.T) {
output := `Instantaneous power reading: 167 Watts
Minimum during sampling period: 124 Watts
Maximum during sampling period: 422 Watts
Average power reading over sample period: 156 Watts
IPMI timestamp: Mon Aug 1 21:22:51 2016
Sampling period: 00699043 Seconds.
Power reading state is: activated
`
expected := []telegraf.Metric{
testutil.MustMetric("ipmi_sensor",
map[string]string{
"name": "instantaneous_power_reading",
"server": "host",
"unit": "watts",
},
map[string]interface{}{"value": float64(167)},
time.Unix(0, 0),
),
testutil.MustMetric("ipmi_sensor",
map[string]string{
"name": "minimum_during_sampling_period",
"server": "host",
"unit": "watts",
},
map[string]interface{}{"value": float64(124)},
time.Unix(0, 0),
),
testutil.MustMetric("ipmi_sensor",
map[string]string{
"name": "maximum_during_sampling_period",
"server": "host",
"unit": "watts",
},
map[string]interface{}{"value": float64(422)},
time.Unix(0, 0),
),
testutil.MustMetric("ipmi_sensor",
map[string]string{
"name": "average_power_reading_over_sample_period",
"server": "host",
"unit": "watts",
},
map[string]interface{}{"value": float64(156)},
time.Unix(0, 0),
),
}
ipmi := &Ipmi{
Log: testutil.Logger{},
}
var acc testutil.Accumulator
err := ipmi.parseDCMIPowerReading(&acc, "host", []byte(output), time.Now())
require.NoError(t, err)
testutil.RequireMetricsEqual(t, expected, acc.GetTelegrafMetrics(), testutil.IgnoreTime())
}
func TestSanitizeIPMICmd(t *testing.T) {
tests := []struct {
name string

48
plugins/inputs/ipmi_sensor/sample.conf
View File

@ -1,41 +1,47 @@
# Read metrics from the bare metal servers via IPMI
[[inputs.ipmi_sensor]]
## optionally specify the path to the ipmitool executable
## Specify the path to the ipmitool executable
# path = "/usr/bin/ipmitool"
##
## Use sudo
## Setting 'use_sudo' to true will make use of sudo to run ipmitool.
## Sudo must be configured to allow the telegraf user to run ipmitool
## without a password.
# use_sudo = false
##
## optionally force session privilege level. Can be CALLBACK, USER, OPERATOR, ADMINISTRATOR
# privilege = "ADMINISTRATOR"
##
## optionally specify one or more servers via a url matching
## Servers
## Specify one or more servers via a url. If no servers are specified, local
## machine sensor stats will be queried. Uses the format:
## [username[:password]@][protocol[(address)]]
## e.g.
## root:passwd@lan(127.0.0.1)
##
## if no servers are specified, local machine sensor stats will be queried
##
## e.g. root:passwd@lan(127.0.0.1)
# servers = ["USERID:PASSW0RD@lan(192.168.1.1)"]
## Recommended: use metric 'interval' that is a multiple of 'timeout' to avoid
## gaps or overlap in pulled data
interval = "30s"
## Session privilege level
## Choose from: CALLBACK, USER, OPERATOR, ADMINISTRATOR
# privilege = "ADMINISTRATOR"
## Timeout for the ipmitool command to complete. Default is 20 seconds.
timeout = "20s"
## Timeout
## Timeout for the ipmitool command to complete.
# timeout = "20s"
## Schema Version: (Optional, defaults to version 1)
metric_version = 2
## Metric schema version
## See the plugin readme for more information on schema versioning.
# metric_version = 1
## Sensors to collect
## Choose from:
## * sdr: default, collects sensor data records
## * chassis_power_status: collects the power status of the chassis
## * dcmi_power_reading: collects the power readings from the Data Center Management Interface
# sensors = ["sdr"]
## Hex key
## Optionally provide the hex key for the IMPI connection.
# hex_key = ""
## Cache
## If ipmitool should use a cache
## for me ipmitool runs about 2 to 10 times faster with cache enabled on HP G10 servers (when using ubuntu20.04)
## the cache file may not work well for you if some sensors come up late
## Using a cache can speed up collection times depending on your device.
# use_cache = false
## Path to the ipmitools cache file (defaults to OS temp dir)