feat(inputs.procstat): Report consistent I/O on Linux (#15186)

CHANGELOG.md

@@ -1,6 +1,16 @@
 <!-- markdownlint-disable MD024 -->
 # Changelog
 
+## Unreleased
+
+### Important Changes
+
+- [PR #15186](https://github.com/influxdata/telegraf/pull/15186) changes the
+  meaning of `inputs.procstat` fields `read_bytes` and `write_bytes` on Linux
+  to now contain _all_ I/O operations for consistency with other
+  operating-systems. The previous values are output as `disk_read_bytes` and
+  `disk_write_bytes` measuring _only_ the I/O on the storage layer.
+
 ## v1.30.2 [2024-04-22]
 
 ### Important Changes

plugins/inputs/procstat/README.md

@@ -159,6 +159,8 @@ Below are an example set of tags and fields:
     - cpu_time_system (float)
     - cpu_time_user (float)
     - cpu_usage (float)
+    - disk_read_bytes (int, Linux only, *telegraf* may need to be ran as **root**)
+    - disk_write_bytes (int, Linux only, *telegraf* may need to be ran as **root**)
     - involuntary_context_switches (int)
     - major_faults (int)
     - memory_anonymous (int)

plugins/inputs/procstat/os_linux.go

@@ -6,6 +6,9 @@ import (
 	"context"
 	"errors"
 	"fmt"
+	"os"
+
+	"github.com/prometheus/procfs"
 
 	"github.com/coreos/go-systemd/v22/dbus"
 	"github.com/shirou/gopsutil/v3/process"
@@ -76,3 +79,27 @@ func findBySystemdUnits(units []string) ([]processGroup, error) {
 func findByWindowsServices(_ []string) ([]processGroup, error) {
 	return nil, nil
 }
+
+func collectTotalReadWrite(proc Process) (r, w uint64, err error) {
+	path := procfs.DefaultMountPoint
+	if hp := os.Getenv("HOST_PROC"); hp != "" {
+		path = hp
+	}
+
+	fs, err := procfs.NewFS(path)
+	if err != nil {
+		return 0, 0, err
+	}
+
+	p, err := fs.Proc(int(proc.PID()))
+	if err != nil {
+		return 0, 0, err
+	}
+
+	stat, err := p.IO()
+	if err != nil {
+		return 0, 0, err
+	}
+
+	return stat.RChar, stat.WChar, nil
+}

plugins/inputs/procstat/os_others.go

@@ -25,3 +25,7 @@ func findBySystemdUnits(_ []string) ([]processGroup, error) {
 func findByWindowsServices(_ []string) ([]processGroup, error) {
 	return nil, nil
 }
+
+func collectTotalReadWrite(_ Process) (r, w uint64, err error) {
+	return 0, 0, errors.ErrUnsupported
+}

plugins/inputs/procstat/os_windows.go

@@ -82,3 +82,7 @@ func findByWindowsServices(services []string) ([]processGroup, error) {
 
 	return groups, nil
 }
+
+func collectTotalReadWrite(_ Process) (r, w uint64, err error) {
+	return 0, 0, errors.ErrUnsupported
+}

plugins/inputs/procstat/process.go

@@ -104,6 +104,15 @@ func (p *Proc) Metric(prefix string, tagging map[string]bool, solarisMode bool)
 		fields[prefix+"write_bytes"] = io.WriteBytes
 	}
 
+	// Linux fixup for gopsutils exposing the disk-only-IO instead of the total
+	// I/O as for example on Windows
+	if rc, wc, err := collectTotalReadWrite(p); err == nil {
+		fields[prefix+"read_bytes"] = rc
+		fields[prefix+"write_bytes"] = wc
+		fields[prefix+"disk_read_bytes"] = io.ReadBytes
+		fields[prefix+"disk_write_bytes"] = io.WriteBytes
+	}
+
 	createdAt, err := p.CreateTime() // returns epoch in ms
 	if err == nil {
 		fields[prefix+"created_at"] = createdAt * 1000000 // ms to ns

plugins/inputs/procstat/procstat_test.go

@@ -347,8 +347,6 @@ func TestGather_ProcessName(t *testing.T) {
 
 	var acc testutil.Accumulator
 	require.NoError(t, p.Gather(&acc))
-
-	testutil.PrintMetrics(acc.GetTelegrafMetrics())
 	require.Equal(t, "custom_name", acc.TagValue("procstat", "process_name"))
 	testutil.RequireMetricsEqual(t, expected, acc.GetTelegrafMetrics(), testutil.IgnoreTime())
 }