feat: sample terminal business latency

This commit is contained in:
2026-09-19 16:30:20 +08:00
parent 8ce9f4e1fa
commit 5a49dd5644
14 changed files with 139 additions and 21 deletions

View File

@@ -29,6 +29,14 @@ type OperationalSnapshot struct {
Truncated bool
// RecentEvents 是固定事件类别在最近窗口内的有界数量。
RecentEvents map[string]uint64
// TaskTerminalLatencySeconds 是最近窗口终态任务从创建到完成的平均耗时样本。
TaskTerminalLatencySeconds float64
// TaskTerminalLatencyMaxSeconds 是最近窗口终态任务样本的最大耗时。
TaskTerminalLatencyMaxSeconds float64
// WorkflowTerminalLatencySeconds 是最近窗口终态流程从创建到完成的平均耗时样本。
WorkflowTerminalLatencySeconds float64
// WorkflowTerminalLatencyMaxSeconds 是最近窗口终态流程样本的最大耗时。
WorkflowTerminalLatencyMaxSeconds float64
}
var (
@@ -76,12 +84,22 @@ func ReadOperationalSnapshot(ctx context.Context, db *sql.DB) (OperationalSnapsh
if err != nil {
return OperationalSnapshot{}, err
}
taskLatency, err := sampledTerminalLatency(ctx, db, "pulse_tasks", "ix_task_observe_terminal_completed", []string{"SUCCEEDED", "FAILED", "CANCELLED"}, now.Add(-operationalMetricEventWindow))
if err != nil {
return OperationalSnapshot{}, err
}
workflowLatency, err := sampledTerminalLatency(ctx, db, "weave_instances", "ix_workflow_observe_terminal_completed", []string{"SUCCEEDED", "FAILED", "CANCELLED"}, now.Add(-operationalMetricEventWindow))
if err != nil {
return OperationalSnapshot{}, err
}
return OperationalSnapshot{
TaskStates: tasks, TaskOldestAgeSeconds: taskAge,
DeliveryStates: deliveries, DeliveryOldestAgeSeconds: deliveryAge,
WorkflowStates: workflows, WorkflowOldestAgeSeconds: workflowAge,
RecentEvents: events,
Truncated: taskTruncated || deliveryTruncated || workflowTruncated || eventTruncated,
RecentEvents: events,
TaskTerminalLatencySeconds: taskLatency.average, TaskTerminalLatencyMaxSeconds: taskLatency.maximum,
WorkflowTerminalLatencySeconds: workflowLatency.average, WorkflowTerminalLatencyMaxSeconds: workflowLatency.maximum,
Truncated: taskTruncated || deliveryTruncated || workflowTruncated || eventTruncated || taskLatency.truncated || workflowLatency.truncated,
}, nil
}
@@ -130,6 +148,58 @@ func boundedRecentEventCounts(ctx context.Context, db *sql.DB, now time.Time) (m
return counts, truncated, nil
}
type terminalLatencySample struct {
average float64
maximum float64
truncated bool
}
func sampledTerminalLatency(ctx context.Context, db *sql.DB, table, index string, states []string, cutoff time.Time) (terminalLatencySample, error) {
placeholders := "?"
args := make([]any, 0, len(states)+2)
args = append(args, states[0])
for _, state := range states[1:] {
placeholders += ",?"
args = append(args, state)
}
args = append(args, cutoff, operationalMetricSampleLimit+1)
query := "SELECT created_at,completed_at FROM " + table + " FORCE INDEX (" + index + ") WHERE state IN (" + placeholders + ") AND completed_at>=? ORDER BY completed_at,id LIMIT ?"
rows, err := db.QueryContext(ctx, query, args...)
if err != nil {
return terminalLatencySample{}, err
}
defer rows.Close()
var total, maximum float64
var read, valid int
for rows.Next() {
read++
if read > operationalMetricSampleLimit {
continue
}
var created, completed time.Time
if err := rows.Scan(&created, &completed); err != nil {
return terminalLatencySample{}, err
}
seconds := completed.Sub(created).Seconds()
if seconds < 0 {
seconds = 0
}
total += seconds
if seconds > maximum {
maximum = seconds
}
valid++
}
if err := rows.Err(); err != nil {
return terminalLatencySample{}, err
}
result := terminalLatencySample{maximum: maximum, truncated: read > operationalMetricSampleLimit}
if valid > 0 {
result.average = total / float64(valid)
}
return result, nil
}
func oldestAge(ctx context.Context, db *sql.DB, table, index string, states []string, column string, now time.Time) (float64, error) {
placeholders := "?"
args := make([]any, 0, len(states))