Files
easyai-ai-gateway/apps/api/internal/store/binary_result_backfill.go
T
wangbo 4f163ea6d7 perf(storage): 拦截任务二进制并本地暂存结果
原因:任务标准结果中的 Base64、Data URI 和 Buffer 会进入 PostgreSQL JSON,导致 TOAST 与备份体积快速增长。

影响:新增统一 JSON 持久化关口;upload_none 将二进制原子写入本地结果目录,数据库仅保存带 SHA-256 的有界占位符;任务详情、同步响应、异步查询和兼容协议按需校验恢复。补充 24 小时清理、容量上限、历史小批量治理命令及管理端说明。

风险:本地结果超过 TTL、丢失或损坏时分别返回明确的 410/500;空间不足时返回 503 且不重试上游。未自动执行历史治理。

验证:三种真实图片模型同步/异步与幂等重放通过;Go vet/全量测试、前端 111 测试、lint/typecheck/build、OpenAPI、迁移安全、govulncheck、依赖审计、手工发布测试及 Linux amd64 构建通过。
2026-07-24 21:13:09 +08:00

88 lines
2.3 KiB
Go

package store
import (
"context"
"encoding/json"
"time"
"github.com/jackc/pgx/v5"
)
type TaskBinaryResultBackfillItem struct {
ID string
Result map[string]any
FinishedAt time.Time
}
func (s *Store) ListTaskBinaryResultBackfillBatch(ctx context.Context, afterID string, batchSize int) ([]TaskBinaryResultBackfillItem, error) {
if batchSize < 1 || batchSize > 100 {
batchSize = 100
}
var items []TaskBinaryResultBackfillItem
err := pgx.BeginFunc(ctx, s.pool, func(tx pgx.Tx) error {
if _, err := tx.Exec(ctx, `SET LOCAL lock_timeout = '5s'; SET LOCAL statement_timeout = '5s'`); err != nil {
return err
}
rows, err := tx.Query(ctx, `
SELECT id::text, result, COALESCE(finished_at, updated_at)
FROM gateway_tasks
WHERE status = 'succeeded'
AND result <> '{}'::jsonb
AND (NULLIF($1::text, '') IS NULL OR id > NULLIF($1::text, '')::uuid)
ORDER BY id
LIMIT $2`, afterID, batchSize)
if err != nil {
return err
}
defer rows.Close()
items = make([]TaskBinaryResultBackfillItem, 0, batchSize)
for rows.Next() {
var item TaskBinaryResultBackfillItem
var resultJSON []byte
if err := rows.Scan(&item.ID, &resultJSON, &item.FinishedAt); err != nil {
return err
}
item.Result = decodeObject(resultJSON)
items = append(items, item)
}
return rows.Err()
})
return items, err
}
func (s *Store) UpdateTaskBinaryResultBackfill(ctx context.Context, taskID string, result map[string]any) (bool, error) {
report := sanitizeJSONForStorageWithReport(minimalTaskResult(result))
if report.BinaryCount > 0 {
return false, &taskPayloadBinaryError{
target: ErrTaskResultBinaryNotMaterialized,
code: "result_binary_not_materialized",
count: report.BinaryCount,
}
}
resultJSON, err := json.Marshal(report.Value)
if err != nil {
return false, err
}
updated := false
err = pgx.BeginFunc(ctx, s.pool, func(tx pgx.Tx) error {
if _, err := tx.Exec(ctx, `SET LOCAL lock_timeout = '5s'; SET LOCAL statement_timeout = '5s'`); err != nil {
return err
}
tag, err := tx.Exec(ctx, `
UPDATE gateway_tasks
SET result = $2::jsonb,
updated_at = now()
WHERE id = $1::uuid
AND status = 'succeeded'`,
taskID,
string(resultJSON),
)
if err != nil {
return err
}
updated = tag.RowsAffected() == 1
return nil
})
return updated, err
}