fix: move driver_version into nvidia_smi_gpu_info labels

Signed-off-by: Utku Ozdemir <[email protected]>

internal/exporter/_query-test.txt

@@ -1,3 +1,3 @@
-uuid, name, driver_model.current, driver_model.pending, vbios_version, fan.speed [%], memory.used [MiB]
-GPU-df6e7a7c-7314-46f8-abc4-b88b36dcf3aa, NVIDIA GeForce RTX 2080 SUPER, WDDM, WDDM, 90.04.7a.40.73, 38 %, 575 MiB
-GPU-04757e3e-3077-4e2e-b988-7e2d647b52e9, Some Other GPU, DoesntMatter, DoesntMatter, 1a.2b.3c.4d, 50 %, 1234 MiB
+uuid, name, driver_model.current, driver_model.pending, vbios_version, driver_version, fan.speed [%], memory.used [MiB]
+GPU-df6e7a7c-7314-46f8-abc4-b88b36dcf3aa, NVIDIA GeForce RTX 2080 SUPER, WDDM, WDDM, 90.04.7a.40.73, 466.63, 38 %, 575 MiB
+GPU-04757e3e-3077-4e2e-b988-7e2d647b52e9, Some Other GPU, DoesntMatter, DoesntMatter, 1a.2b.3c.4d, 123.45, 50 %, 1234 MiB

internal/exporter/csv.go

@@ -1,6 +1,9 @@
 package exporter
 
-import "strings"
+import (
+	"fmt"
+	"strings"
+)
 
 type table struct {
 	rows          []row
@@ -19,7 +22,7 @@ type cell struct {
 	rawValue string
 }
 
-func parseCSVIntoTable(queryResult string, qFields []qField) table {
+func parseCSVIntoTable(queryResult string, qFields []qField) (table, error) {
 	lines := strings.Split(strings.TrimSpace(queryResult), "\n")
 	titlesLine := lines[0]
 	valuesLines := lines[1:]
@@ -39,6 +42,10 @@ func parseCSVIntoTable(queryResult string, qFields []qField) table {
 		qFieldToCell := make(map[qField]cell, numCols)
 		cells := make([]cell, numCols)
 		rawValues := parseCSVLine(valuesLine)
+		if len(qFields) != len(rFields) {
+			return table{}, fmt.Errorf("query fields (%d) and returned fields (%d) have different sizes", len(qFields), len(rFields))
+		}
+
 		for colIndex, rawValue := range rawValues {
 			q := qFields[colIndex]
 			r := rFields[colIndex]
@@ -65,7 +72,7 @@ func parseCSVIntoTable(queryResult string, qFields []qField) table {
 		rows:          rows,
 		rFields:       rFields,
 		qFieldToCells: qFieldToCells,
-	}
+	}, nil
 }
 
 func parseCSVLine(line string) []string {

internal/exporter/csv_test.go

@@ -14,7 +14,8 @@ Some Dummy GPU, 12.34 W
 )
 
 func TestParseCsvIntoTable(t *testing.T) {
-	parsed := parseCSVIntoTable(testCsv, []qField{"name", "power.draw"})
+	parsed, err := parseCSVIntoTable(testCsv, []qField{"name", "power.draw"})
+	assert.NoError(t, err)
 	assert.Len(t, parsed.rows, 2)
 	assert.Equal(t, []rField{"name", "power.draw [W]"}, parsed.rFields)
 

internal/exporter/exporter.go

@@ -33,6 +33,7 @@ var (
 		{qField: driverModelCurrentQField, label: "driver_model_current"},
 		{qField: driverModelPendingQField, label: "driver_model_pending"},
 		{qField: vBiosVersionQField, label: "vbios_version"},
+		{qField: driverVersionQField, label: "driver_version"},
 	}
 
 	runCmd = func(cmd *exec.Cmd) error { return cmd.Run() }
@@ -156,9 +157,11 @@ func (e *gpuExporter) Collect(ch chan<- prometheus.Metric) {
 		driverModelCurrent := r.qFieldToCells[driverModelCurrentQField].rawValue
 		driverModelPending := r.qFieldToCells[driverModelPendingQField].rawValue
 		vBiosVersion := r.qFieldToCells[vBiosVersionQField].rawValue
+		driverVersion := r.qFieldToCells[driverVersionQField].rawValue
 
 		infoMetric := prometheus.MustNewConstMetric(e.gpuInfoDesc, prometheus.GaugeValue,
-			1, uuid, name, driverModelCurrent, driverModelPending, vBiosVersion)
+			1, uuid, name, driverModelCurrent,
+			driverModelPending, vBiosVersion, driverVersion)
 		ch <- infoMetric
 
 		for _, c := range r.cells {
@@ -193,7 +196,11 @@ func scrape(qFields []qField, nvidiaSmiCommand string) (*table, error) {
 		return nil, fmt.Errorf("command failed. stderr: %s err: %w", stderr.String(), err)
 	}
 
-	t := parseCSVIntoTable(strings.TrimSpace(stdout.String()), qFields)
+	t, err := parseCSVIntoTable(strings.TrimSpace(stdout.String()), qFields)
+	if err != nil {
+		return nil, err
+	}
+
 	return &t, nil
 }
 

internal/exporter/exporter_test.go

@@ -157,7 +157,7 @@ end:
 		}
 	}
 
-	assert.Len(t, descStrs, 9)
+	assert.Len(t, descStrs, 10)
 	descs := strings.Join(descStrs, "\n")
 	assert.Contains(t, descs, "aaa_fan_speed")
 	assert.Contains(t, descs, "aaa_memory_used")
@@ -168,6 +168,7 @@ end:
 	assert.Contains(t, descs, "aaa_driver_model_current")
 	assert.Contains(t, descs, "aaa_driver_model_pending")
 	assert.Contains(t, descs, "aaa_vbios_version")
+	assert.Contains(t, descs, "aaa_driver_version")
 }
 
 func TestCollect(t *testing.T) {
@@ -182,7 +183,7 @@ func TestCollect(t *testing.T) {
 	logger := log.NewNopLogger()
 	exp, err := New("aaa", "bbb",
 		"uuid,name,driver_model.current,driver_model.pending,"+
-			"vbios_version,fan.speed,memory.used", logger)
+			"vbios_version,driver_version,fan.speed,memory.used", logger)
 	assert.NoError(t, err)
 
 	doneCh := make(chan bool)
@@ -207,7 +208,7 @@ end:
 
 	metricsJoined := strings.Join(metrics, "\n")
 
-	assert.Len(t, metrics, 7)
+	assert.Len(t, metrics, 9)
 	assert.Contains(t, metricsJoined, "aaa_gpu_info")
 	assert.Contains(t, metricsJoined, "aaa_name")
 	assert.Contains(t, metricsJoined, "aaa_fan_speed_ratio")

internal/exporter/fields.go

@@ -2,6 +2,7 @@ package exporter
 
 import (
 	"bytes"
+	"errors"
 	"os/exec"
 	"regexp"
 	"strings"
@@ -13,6 +14,7 @@ const (
 	driverModelCurrentQField qField = "driver_model.current"
 	driverModelPendingQField qField = "driver_model.pending"
 	vBiosVersionQField       qField = "vbios_version"
+	driverVersionQField      qField = "driver_version"
 	qFieldsAuto                     = "AUTO"
 	DefaultQField                   = qFieldsAuto
 )
@@ -153,6 +155,9 @@ func ParseAutoQFields(nvidiaSmiCommand string) ([]qField, error) {
 
 	out := stdout.String()
 	fields := extractQFields(out)
+	if fields == nil {
+		return nil, errors.New("could not extract any query fields")
+	}
 	return fields, nil
 }
 

samples/sample-source.sh

@@ -1,9 +1,4 @@
 #!/usr/bin/env sh
 
-
-
-
-echo "timestamp, driver_version, count, name, serial, uuid, pci.bus_id, pci.domain, pci.bus, pci.device, pci.device_id, pci.sub_device_id, pcie.link.gen.current, pcie.link.gen.max, pcie.link.width.current, pcie.link.width.max, index, display_mode, display_active, persistence_mode, accounting.mode, accounting.buffer_size, driver_model.current, driver_model.pending, vbios_version, inforom.img, inforom.oem, inforom.ecc, inforom.pwr, gom.current, gom.pending, fan.speed [%], pstate, clocks_throttle_reasons.supported, clocks_throttle_reasons.active, clocks_throttle_reasons.gpu_idle, clocks_throttle_reasons.applications_clocks_setting, clocks_throttle_reasons.sw_power_cap, clocks_throttle_reasons.hw_slowdown, clocks_throttle_reasons.hw_thermal_slowdown, clocks_throttle_reasons.hw_power_brake_slowdown, clocks_throttle_reasons.sw_thermal_slowdown, clocks_throttle_reasons.sync_boost, memory.total [MiB], memory.used [MiB], memory.free [MiB], compute_mode, utilization.gpu [%], utilization.memory [%], encoder.stats.sessionCount, encoder.stats.averageFps, encoder.stats.averageLatency, ecc.mode.current, ecc.mode.pending, ecc.errors.corrected.volatile.device_memory, ecc.errors.corrected.volatile.dram, ecc.errors.corrected.volatile.register_file, ecc.errors.corrected.volatile.l1_cache, ecc.errors.corrected.volatile.l2_cache, ecc.errors.corrected.volatile.texture_memory, ecc.errors.corrected.volatile.cbu, ecc.errors.corrected.volatile.sram, ecc.errors.corrected.volatile.total, ecc.errors.corrected.aggregate.device_memory, ecc.errors.corrected.aggregate.dram, ecc.errors.corrected.aggregate.register_file, ecc.errors.corrected.aggregate.l1_cache, ecc.errors.corrected.aggregate.l2_cache, ecc.errors.corrected.aggregate.texture_memory, ecc.errors.corrected.aggregate.cbu, ecc.errors.corrected.aggregate.sram, ecc.errors.corrected.aggregate.total, ecc.errors.uncorrected.volatile.device_memory, ecc.errors.uncorrected.volatile.dram, ecc.errors.uncorrected.volatile.register_file, ecc.errors.uncorrected.volatile.l1_cache, ecc.errors.uncorrected.volatile.l2_cache, ecc.errors.uncorrected.volatile.texture_memory, ecc.errors.uncorrected.volatile.cbu, ecc.errors.uncorrected.volatile.sram, ecc.errors.uncorrected.volatile.total, ecc.errors.uncorrected.aggregate.device_memory, ecc.errors.uncorrected.aggregate.dram, ecc.errors.uncorrected.aggregate.register_file, ecc.errors.uncorrected.aggregate.l1_cache, ecc.errors.uncorrected.aggregate.l2_cache, ecc.errors.uncorrected.aggregate.texture_memory, ecc.errors.uncorrected.aggregate.cbu, ecc.errors.uncorrected.aggregate.sram, ecc.errors.uncorrected.aggregate.total, retired_pages.single_bit_ecc.count, retired_pages.double_bit.count, retired_pages.pending, temperature.gpu, temperature.memory, power.management, power.draw [W], power.limit [W], enforced.power.limit [W], power.default_limit [W], power.min_limit [W], power.max_limit [W], clocks.current.graphics [MHz], clocks.current.sm [MHz], clocks.current.memory [MHz], clocks.current.video [MHz], clocks.applications.graphics [MHz], clocks.applications.memory [MHz], clocks.default_applications.graphics [MHz], clocks.default_applications.memory [MHz], clocks.max.graphics [MHz], clocks.max.sm [MHz], clocks.max.memory [MHz], mig.mode.current, mig.mode.pending"
-echo "2021/06/09 23:27:38.358, 466.63, 1, NVIDIA GeForce RTX 2080 SUPER, [N/A], GPU-df6e7a7c-7314-46f8-abc4-b88b36dcf3aa, 00000000:0C:00.0, 0x0000, 0x0C, 0x00, 0x1E8110DE, 0x40051458, 1, 3, 16, 16, 0, Enabled, Enabled, [N/A], Disabled, 4000, WDDM, WDDM, 90.04.7A.40.73, G001.0000.02.04, 1.1, [N/A], [N/A], [N/A], [N/A], 38 %, P8, 0x00000000000001FF, 0x0000000000000001, Active, Not Active, Not Active, Not Active, Not Active, Not Active, Not Active, Not Active, 8192 MiB, 856 MiB, 7336 MiB, Default, 0 %, 10 %, 0, 0, 0, [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], 35, N/A, Enabled, 31.12 W, 250.00 W, 250.00 W, 250.00 W, 105.00 W, 350.00 W, 300 MHz, 300 MHz, 405 MHz, 540 MHz, [N/A], [N/A], [N/A], [N/A], 2145 MHz, 2145 MHz, 7751 MHz, [N/A], [N/A]"
-echo "2042/06/09 23:27:38.358, 123.45, 1, Dummy GPU 1, [N/A], GPU-523e933c-cf12-4d85-95b4-dd144e8fc516, 00000000:0C:00.0, 0x0000, 0x0C, 0x00, 0x1E8110DE, 0x40051458, 1, 3, 16, 16, 0, Enabled, Enabled, [N/A], Disabled, 4000, WDDM, WDDM, 90.04.7A.40.73, G001.0000.02.04, 1.1, [N/A], [N/A], [N/A], [N/A], 38 %, P8, 0x00000000000001FF, 0x0000000000000001, Active, Not Active, Not Active, Not Active, Not Active, Not Active, Not Active, Not Active, 8192 MiB, 856 MiB, 7336 MiB, Default, 0 %, 10 %, 0, 0, 0, [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], 35, N/A, Enabled, 31.12 W, 250.00 W, 250.00 W, 250.00 W, 105.00 W, 350.00 W, 300 MHz, 300 MHz, 405 MHz, 540 MHz, [N/A], [N/A], [N/A], [N/A], 2145 MHz, 2145 MHz, 7751 MHz, [N/A], [N/A]"
-echo "2021/06/12 18:45:18.358, 466.63, 1, Dummy GPU 2, [N/A], GPU-b2fe4f12-c3dd-4fa4-914b-9e7b975a0faa, 00000000:0C:00.0, 0x0000, 0x0C, 0x00, 0x1E8110DE, 0x40051458, 1, 3, 16, 16, 0, Enabled, Disabled, [N/A], Disabled, 4000, WDDM, WDDM, 90.04.7A.40.73, G001.0000.02.04, 1.1, [N/A], [N/A], [N/A], [N/A], 38 %, P8, 0x00000000000001FF, 0x0000000000000001, Active, Not Active, Not Active, Not Active, Not Active, Not Active, Not Active, Not Active, 8192 MiB, 779 MiB, 7413 MiB, Default, 2 %, 0 %, 0, 0, 0, [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], [N/A], 34, N/A, Enabled, 29.50 W, 250.00 W, 250.00 W, 250.00 W, 105.00 W, 350.00 W, 300 MHz, 300 MHz, 405 MHz, 540 MHz, [N/A], [N/A], [N/A], [N/A], 2235 MHz, 2235 MHz, 7751 MHz, [N/A], [N/A]"
+echo "driver_version,uuid,name,driver_model.current,driver_model.pending,vbios_version"
+echo "460.91.03,GPU-df6e7a7c-7314-46f8-abc4-b88b36dcf3aa,NVIDIA GeForce RTX 2080 SUPER, WDDM, WDDM, 90.04.7A.40.73"