每日清理

This commit is contained in:
2026-08-20 12:03:17 +08:00
parent 143df60485
commit 22b95db2f9
23 changed files with 614 additions and 581 deletions

View File

@@ -13,75 +13,108 @@ import (
"github.com/break/junhong_cmp_fiber/pkg/errors"
)
// Source 表示受在线留存边界约束的数据源。
type Source string
const (
// SourceAudit 表示统一审计事件。
SourceAudit Source = constants.AuditArchiveSource
// SourceIntegration 表示外部交互日志。
SourceAudit Source = constants.AuditArchiveSource
SourceIntegration Source = constants.IntegrationArchiveSource
)
// Info 是查询响应公开的在线留存边界。
type Info struct {
OnlineFrom time.Time `json:"online_from" description:"当前可在线查询的最早时间"`
ArchivedBefore *time.Time `json:"archived_before" description:"早于该时间的数据已归档;尚未清理时为空"`
Timezone string `json:"timezone" description:"留存自然日时区"`
}
// Load 从归档账本读取已完成物理清理的数据边界
// Load 仅公开从最早在线日期开始连续完成物理清理的边界,绝不跨越清理空洞
func Load(ctx context.Context, db *gorm.DB, sources ...Source) (Info, error) {
location, err := time.LoadLocation(constants.AuditArchiveTimezone)
if err != nil {
return Info{}, errors.Wrap(errors.CodeInternalError, err, "加载审计留存时区失败")
}
now := time.Now().In(location)
info := Info{OnlineFrom: time.Date(now.Year(), now.Month(), 1, 0, 0, 0, 0, location), Timezone: constants.AuditArchiveTimezone}
boundaries := make([]sourceRetention, 0, len(sources))
for _, source := range sources {
boundary, cleaned, err := sourceBoundary(ctx, db, source, location)
boundary, err := sourceBoundary(ctx, db, source, location)
if err != nil {
return Info{}, err
}
if boundary.Before(info.OnlineFrom) && info.ArchivedBefore == nil {
info.OnlineFrom = boundary
boundaries = append(boundaries, boundary)
}
now := time.Now().In(location)
fallback := time.Date(now.Year(), now.Month(), now.Day(), 0, 0, 0, 0, location)
info := Info{OnlineFrom: fallback, Timezone: constants.AuditArchiveTimezone}
if len(boundaries) == 0 {
return info, nil
}
for _, boundary := range boundaries {
if boundary.onlineFrom.Before(info.OnlineFrom) {
info.OnlineFrom = boundary.onlineFrom
}
if cleaned && (info.ArchivedBefore == nil || boundary.After(*info.ArchivedBefore)) {
value := boundary
info.ArchivedBefore = &value
info.OnlineFrom = boundary
}
if len(boundaries) == 1 && boundaries[0].cleaned {
value := boundaries[0].onlineFrom
info.ArchivedBefore = &value
return info, nil
}
if len(boundaries) > 1 {
common := boundaries[0].onlineFrom
allCleaned := boundaries[0].cleaned
for _, boundary := range boundaries[1:] {
if boundary.onlineFrom.Before(common) {
common = boundary.onlineFrom
}
allCleaned = allCleaned && boundary.cleaned
}
if allCleaned {
info.OnlineFrom = common
info.ArchivedBefore = &common
}
}
return info, nil
}
func sourceBoundary(ctx context.Context, db *gorm.DB, source Source, location *time.Location) (time.Time, bool, error) {
var cleanedEnd sql.NullTime
if err := db.WithContext(ctx).Model(&model.LogArchiveRun{}).
Where("source = ? AND cleaned_at IS NOT NULL", source).
Select("MAX(range_end)").Scan(&cleanedEnd).Error; err != nil {
return time.Time{}, false, errors.Wrap(errors.CodeDatabaseError, err, "查询审计留存清理边界失败")
}
if cleanedEnd.Valid {
return cleanedEnd.Time.In(location), true, nil
}
type sourceRetention struct {
onlineFrom time.Time
cleaned bool
}
var earliest sql.NullTime
table, column := "tb_audit_event", "occurred_at"
func sourceBoundary(ctx context.Context, db *gorm.DB, source Source, location *time.Location) (sourceRetention, error) {
table, column := "tb_audit_event", "created_at"
if source == SourceIntegration {
table, column = "tb_integration_log", "created_at"
}
if err := db.WithContext(ctx).Table(table).Select("MIN(" + column + ")").Scan(&earliest).Error; err != nil {
return time.Time{}, false, errors.Wrap(errors.CodeDatabaseError, err, "查询审计在线数据边界失败")
var earliestOnline, earliestLedger sql.NullTime
if err := db.WithContext(ctx).Table(table).Select("MIN(" + column + ")").Scan(&earliestOnline).Error; err != nil {
return sourceRetention{}, errors.Wrap(errors.CodeDatabaseError, err, "查询审计在线数据边界失败")
}
if earliest.Valid {
return earliest.Time.In(location), false, nil
if err := db.WithContext(ctx).Model(&model.LogArchiveRun{}).Where("source = ?", source).Select("MIN(archive_date)").Scan(&earliestLedger).Error; err != nil {
return sourceRetention{}, errors.Wrap(errors.CodeDatabaseError, err, "查询审计留存账本边界失败")
}
now := time.Now().In(location)
return time.Date(now.Year(), now.Month(), 1, 0, 0, 0, 0, location), false, nil
if !earliestOnline.Valid && !earliestLedger.Valid {
now := time.Now().In(location)
return sourceRetention{onlineFrom: time.Date(now.Year(), now.Month(), now.Day(), 0, 0, 0, 0, location)}, nil
}
start := earliestLedger.Time.In(location)
if earliestOnline.Valid && (!earliestLedger.Valid || earliestOnline.Time.Before(earliestLedger.Time)) {
start = earliestOnline.Time.In(location)
}
start = time.Date(start.Year(), start.Month(), start.Day(), 0, 0, 0, 0, location)
var rows []model.LogArchiveRun
if err := db.WithContext(ctx).Where("source = ? AND archive_date >= ?", source, start.Format(time.DateOnly)).Order("archive_date ASC").Find(&rows).Error; err != nil {
return sourceRetention{}, errors.Wrap(errors.CodeDatabaseError, err, "查询审计留存清理边界失败")
}
expected := start
for _, row := range rows {
date := row.ArchiveDate
date = time.Date(date.Year(), date.Month(), date.Day(), 0, 0, 0, 0, location)
if !date.Equal(expected) || row.CleanedAt == nil {
break
}
expected = expected.AddDate(0, 0, 1)
}
return sourceRetention{onlineFrom: expected, cleaned: !expected.Equal(start)}, nil
}
// NormalizeRange 将缺省范围收敛到在线窗口,并拒绝归档或跨边界查询。
func NormalizeRange(info Info, from, to *time.Time, maxRange ...time.Duration) (*time.Time, *time.Time, error) {
explicitFrom := from != nil
if from != nil && info.ArchivedBefore != nil && from.Before(info.OnlineFrom) {
@@ -110,7 +143,6 @@ func NormalizeRange(info Info, from, to *time.Time, maxRange ...time.Duration) (
}
return from, to, nil
}
func archivedError(info Info) error {
return errors.NewWithData(errors.CodeAuditDataArchived, map[string]any{"retention": info})
}