feat(system-metrics): update metrics to gather more information

system_metrics.go

@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"log/slog"
 	"slices"
+	"sort"
 	"time"
 
 	"github.com/elastic/go-elasticsearch/v7"
@@ -13,6 +14,7 @@ import (
 	"github.com/shirou/gopsutil/host"
 	"github.com/shirou/gopsutil/mem"
 	"github.com/shirou/gopsutil/net"
+	"github.com/shirou/gopsutil/process"
 )
 
 type SystemMetricsCollector struct {
@@ -81,9 +83,79 @@ func (smc *SystemMetricsCollector) collectMetrics() (SystemResources, error) {
 		}
 	}
 
+	if smc.config.CollectProcesses {
+		if err := smc.collectProcessMetrics(&result); err != nil {
+			slog.Warn("failed to collect process metrics", "error", err)
+		}
+	}
+
 	return result, nil
 }
 
+func (smc *SystemMetricsCollector) collectProcessMetrics(result *SystemResources) error {
+	processes, err := process.Processes()
+	if err != nil {
+		return err
+	}
+
+	var processInfos []ProcessInfo
+	var totalOpenFiles int32
+
+	for _, p := range processes {
+		name, err := p.Name()
+		if err != nil {
+			continue
+		}
+
+		cpuPercent, err := p.CPUPercent()
+		if err != nil {
+			continue
+		}
+
+		memInfo, err := p.MemoryInfo()
+		if err != nil {
+			continue
+		}
+
+		status, err := p.Status()
+		if err != nil {
+			status = ""
+		}
+
+		createTime, err := p.CreateTime()
+		if err != nil {
+			createTime = 0
+		}
+
+		if openFiles, err := p.NumFDs(); err == nil {
+			totalOpenFiles += openFiles
+		}
+
+		processInfos = append(processInfos, ProcessInfo{
+			PID:        p.Pid,
+			Name:       name,
+			CPUPercent: cpuPercent,
+			MemoryMB:   float32(memInfo.RSS) / 1024 / 1024,
+			Status:     status,
+			CreateTime: createTime,
+		})
+	}
+
+	sort.Slice(processInfos, func(i, j int) bool {
+		return processInfos[i].CPUPercent > processInfos[j].CPUPercent
+	})
+
+	limit := smc.config.TopProcessesLimit
+	if len(processInfos) > limit {
+		processInfos = processInfos[:limit]
+	}
+
+	result.TopProcesses = processInfos
+	result.OpenFileDescriptors = totalOpenFiles
+
+	return nil
+}
+
 func (smc *SystemMetricsCollector) collectCPUMetrics(result *SystemResources) error {
 	cpuPercents, err := cpu.Percent(time.Second, false)
 	if err != nil {