feat(inputs.system): Add option to specify collected information

plugins/inputs/system/README.md

@@ -20,7 +20,12 @@ plugin ordering. See [CONFIGURATION.md][CONFIGURATION.md] for more details.
 ```toml @sample.conf
 # Read metrics about system load & uptime
 [[inputs.system]]
-  # no configuration
+  ## Metric groups to collect.
+  ## Available options:
+  ##   load   - system gauge metrics (load averages, cpu counts, user counts)
+  ##   uptime - system uptime
+  ## By default all groups are collected.
+  # collect = ["load", "uptime"]
 ```
 
 ### Permissions
@@ -35,22 +40,70 @@ same requirements for `n_users` apply.
 
 ## Metrics
 
-- system
-  - fields:
-    - load1 (float)
-    - load15 (float)
-    - load5 (float)
-    - n_users (integer)
-    - n_unique_users (integer)
-    - n_cpus (integer)
-    - n_physical_cpus (integer)
-    - uptime (integer, seconds)
-    - uptime_format (string, deprecated in 1.10, use `uptime` field)
+### `system`
+
+All fields below belong to the `system` measurement. The `collect` option
+controls which groups are gathered.
+
+| Field             | Group    | Type    | Description                                    |
+|-------------------|----------|---------|------------------------------------------------|
+| `load1`           | `load`   | float   | 1-minute load average                          |
+| `load5`           | `load`   | float   | 5-minute load average                          |
+| `load15`          | `load`   | float   | 15-minute load average                         |
+| `n_users`         | `load`   | integer | Number of logged-in user sessions              |
+| `n_unique_users`  | `load`   | integer | Number of unique logged-in usernames           |
+| `n_cpus`          | `load`   | integer | Number of logical CPUs                         |
+| `n_physical_cpus` | `load`   | integer | Number of physical CPUs                        |
+| `uptime`          | `uptime` | integer | System uptime in seconds                       |
+| `uptime_format`   | `uptime` | string  | Human-readable uptime (deprecated, use uptime) |
 
 ## Example Output
 
 ```text
-system,host=tyrion load1=3.72,load5=2.4,load15=2.1,n_users=3i,n_cpus=4i,n_physical_cpus=2i 1483964144000000000
-system,host=tyrion uptime=1249632i 1483964144000000000
-system,host=tyrion uptime_format="14 days, 11:07" 1483964144000000000
+system,host=worker-01 load1=3.72,load5=2.4,load15=2.1,n_users=3i,n_unique_users=2i,n_cpus=4i,n_physical_cpus=2i 1748000000000000000
+system,host=worker-01 uptime=1249632i 1748000000000000000
+system,host=worker-01 uptime_format="14 days, 11:07" 1748000000000000000
+```
+
+## Example Output (Prometheus)
+
+When using the [Prometheus output plugin][prom-output] or
+[Prometheus client plugin][prom-client], Telegraf converts each field into
+its own Prometheus metric by appending the field name to the measurement name.
+
+[prom-output]: ../../../plugins/outputs/prometheus_client/README.md
+[prom-client]: ../../../plugins/outputs/prometheus_client/README.md
+
+```text
+# HELP system_load1 Telegraf collected metric
+# TYPE system_load1 gauge
+system_load1{host="worker-01"} 3.72
+
+# HELP system_load15 Telegraf collected metric
+# TYPE system_load15 gauge
+system_load15{host="worker-01"} 2.1
+
+# HELP system_load5 Telegraf collected metric
+# TYPE system_load5 gauge
+system_load5{host="worker-01"} 2.4
+
+# HELP system_n_cpus Telegraf collected metric
+# TYPE system_n_cpus gauge
+system_n_cpus{host="worker-01"} 4
+
+# HELP system_n_physical_cpus Telegraf collected metric
+# TYPE system_n_physical_cpus gauge
+system_n_physical_cpus{host="worker-01"} 2
+
+# HELP system_n_unique_users Telegraf collected metric
+# TYPE system_n_unique_users gauge
+system_n_unique_users{host="worker-01"} 2
+
+# HELP system_n_users Telegraf collected metric
+# TYPE system_n_users gauge
+system_n_users{host="worker-01"} 3
+
+# HELP system_uptime Telegraf collected metric
+# TYPE system_uptime counter
+system_uptime{host="worker-01"} 1249632
 ```

plugins/inputs/system/sample.conf

@@ -1,3 +1,8 @@
 # Read metrics about system load & uptime
 [[inputs.system]]
-  # no configuration
+  ## Metric groups to collect.
+  ## Available options:
+  ##   load   - system gauge metrics (load averages, cpu counts, user counts)
+  ##   uptime - system uptime
+  ## By default all groups are collected.
+  # collect = ["load", "uptime"]

plugins/inputs/system/system.go

@@ -15,68 +15,95 @@ import (
 	"github.com/shirou/gopsutil/v4/load"
 
 	"github.com/influxdata/telegraf"
+	"github.com/influxdata/telegraf/internal/choice"
 	"github.com/influxdata/telegraf/plugins/inputs"
 )
 
 //go:embed sample.conf
 var sampleConfig string
 
+var availableCollectors = []string{"load", "uptime"}
+
 type System struct {
+	Collect []string `toml:"collect"`
+
 	Log telegraf.Logger `toml:"-"`
+
+	collectLoad   bool
+	collectUptime bool
 }
 
 func (*System) SampleConfig() string {
 	return sampleConfig
 }
 
-func (s *System) Gather(acc telegraf.Accumulator) error {
-	loadavg, err := load.Avg()
-	if err != nil && !strings.Contains(err.Error(), "not implemented") {
-		return err
+func (s *System) Init() error {
+	if len(s.Collect) == 0 {
+		s.Collect = availableCollectors
 	}
-
-	numLogicalCPUs, err := cpu.Counts(true)
-	if err != nil {
-		return err
+	if err := choice.CheckSlice(s.Collect, availableCollectors); err != nil {
+		return fmt.Errorf("config option 'collect': %w", err)
 	}
 
-	numPhysicalCPUs, err := cpu.Counts(false)
-	if err != nil {
-		return err
-	}
+	s.collectLoad = choice.Contains("load", s.Collect)
+	s.collectUptime = choice.Contains("uptime", s.Collect)
 
-	fields := map[string]interface{}{
-		"load1":           loadavg.Load1,
-		"load5":           loadavg.Load5,
-		"load15":          loadavg.Load15,
-		"n_cpus":          numLogicalCPUs,
-		"n_physical_cpus": numPhysicalCPUs,
-	}
-
-	users, err := host.Users()
-	if err == nil {
-		fields["n_users"] = len(users)
-		fields["n_unique_users"] = findUniqueUsers(users)
-	} else if os.IsNotExist(err) {
-		s.Log.Debugf("Reading users: %s", err.Error())
-	} else if os.IsPermission(err) {
-		s.Log.Debug(err.Error())
-	}
+	return nil
+}
 
+func (s *System) Gather(acc telegraf.Accumulator) error {
 	now := time.Now()
-	acc.AddGauge("system", fields, nil, now)
 
-	uptime, err := host.Uptime()
-	if err != nil {
-		return err
+	if s.collectLoad {
+		fields := make(map[string]interface{})
+
+		loadavg, err := load.Avg()
+		if err != nil {
+			if !strings.Contains(err.Error(), "not implemented") {
+				return err
+			}
+		} else {
+			fields["load1"] = loadavg.Load1
+			fields["load5"] = loadavg.Load5
+			fields["load15"] = loadavg.Load15
+		}
+
+		numLogicalCPUs, err := cpu.Counts(true)
+		if err != nil {
+			return err
+		}
+		numPhysicalCPUs, err := cpu.Counts(false)
+		if err != nil {
+			return err
+		}
+		fields["n_cpus"] = numLogicalCPUs
+		fields["n_physical_cpus"] = numPhysicalCPUs
+
+		users, err := host.Users()
+		if err == nil {
+			fields["n_users"] = len(users)
+			fields["n_unique_users"] = findUniqueUsers(users)
+		} else if os.IsNotExist(err) {
+			s.Log.Debugf("Reading users: %s", err.Error())
+		} else if os.IsPermission(err) {
+			s.Log.Debug(err.Error())
+		}
+
+		acc.AddGauge("system", fields, nil, now)
 	}
 
-	acc.AddCounter("system", map[string]interface{}{
-		"uptime": uptime,
-	}, nil, now)
-	acc.AddFields("system", map[string]interface{}{
-		"uptime_format": formatUptime(uptime),
-	}, nil, now)
+	if s.collectUptime {
+		uptime, err := host.Uptime()
+		if err != nil {
+			return err
+		}
+		acc.AddCounter("system", map[string]interface{}{
+			"uptime": uptime,
+		}, nil, now)
+		acc.AddFields("system", map[string]interface{}{
+			"uptime_format": formatUptime(uptime),
+		}, nil, now)
+	}
 
 	return nil
 }
@@ -88,7 +115,6 @@ func findUniqueUsers(userStats []host.UserStat) int {
 			uniqueUsers[userstat.User] = true
 		}
 	}
-
 	return len(uniqueUsers)
 }
 
@@ -97,7 +123,6 @@ func formatUptime(uptime uint64) string {
 	w := bufio.NewWriter(buf)
 
 	days := uptime / (60 * 60 * 24)
-
 	if days != 0 {
 		s := ""
 		if days > 1 {