DeOldify服务审计日志记录所有图片上传、处理、下载行为与时间戳1. 为什么需要审计日志想象一下你搭建了一个DeOldify黑白图片上色服务每天都有用户上传老照片进行处理。突然有一天有人问“昨天下午3点谁上传了那张全家福处理结果发给谁了”或者“我们这个月总共处理了多少张图片平均处理时间是多少”如果你回答不上来那就尴尬了。这就是审计日志的价值所在。它就像服务的“行车记录仪”完整记录每一次操作的来龙去脉。对于DeOldify这样的图像处理服务审计日志能帮你追踪操作历史谁在什么时候上传了什么图片处理结果给了谁分析使用情况了解服务的使用频率、高峰时段、处理成功率排查问题当处理失败时能快速定位是图片问题、网络问题还是服务问题安全审计监控异常访问防止恶意使用计费依据如果按使用量收费日志就是最准确的计费凭证今天我就带你为DeOldify服务添加一套完整的审计日志系统记录所有关键操作的时间戳和详细信息。2. 审计日志系统设计2.1 记录哪些信息一个好的审计日志应该包含“5W1H”要素Who谁发起的请求IP地址、用户IDWhat做了什么操作上传、处理、下载When什么时候发生的精确到毫秒的时间戳Where从哪里来的请求来源IP、User-AgentWhy操作结果如何成功/失败、错误信息How如何处理的处理时长、输出格式对于DeOldify服务我们需要记录图片上传谁上传了什么图片文件大小、格式、上传时间处理请求处理开始时间、使用的模型、参数设置处理结果处理结束时间、是否成功、处理耗时结果下载谁下载了处理结果、下载时间服务状态服务启动、停止、异常事件2.2 日志存储方案根据使用场景不同可以选择不同的存储方案存储方式优点缺点适用场景文件日志简单、无需额外依赖、易于查看查询复杂、性能一般小规模、单机部署数据库查询灵活、支持复杂分析需要数据库服务、有性能开销中大规模、需要统计分析日志系统专业、支持实时分析、扩展性好架构复杂、学习成本高大规模、生产环境对于大多数DeOldify服务文件日志数据库的组合是比较实用的选择。日常查看用文件日志统计分析用数据库。3. 实现审计日志功能3.1 修改DeOldify服务代码首先我们需要在现有的DeOldify服务中添加日志记录功能。这里以Flask应用为例import logging import json import time from datetime import datetime from flask import Flask, request, jsonify import sqlite3 import os app Flask(__name__) # 配置日志 logging.basicConfig( levellogging.INFO, format%(asctime)s - %(name)s - %(levelname)s - %(message)s, handlers[ logging.FileHandler(deoldify_audit.log), logging.StreamHandler() ] ) logger logging.getLogger(__name__) # 初始化数据库 def init_db(): conn sqlite3.connect(deoldify_audit.db) c conn.cursor() # 创建审计日志表 c.execute( CREATE TABLE IF NOT EXISTS audit_logs ( id INTEGER PRIMARY KEY AUTOINCREMENT, timestamp TEXT NOT NULL, client_ip TEXT, user_agent TEXT, operation TEXT NOT NULL, file_name TEXT, file_size INTEGER, file_type TEXT, request_id TEXT, status TEXT NOT NULL, processing_time REAL, error_message TEXT, additional_info TEXT ) ) # 创建索引以提高查询性能 c.execute(CREATE INDEX IF NOT EXISTS idx_timestamp ON audit_logs(timestamp)) c.execute(CREATE INDEX IF NOT EXISTS idx_operation ON audit_logs(operation)) c.execute(CREATE INDEX IF NOT EXISTS idx_status ON audit_logs(status)) conn.commit() conn.close() # 记录审计日志 def log_audit_event(operation, status, **kwargs): 记录审计日志到文件和数据库 # 获取请求信息 client_ip request.remote_addr if request else N/A user_agent request.headers.get(User-Agent, N/A) if request else N/A # 构建日志记录 log_entry { timestamp: datetime.now().isoformat(), client_ip: client_ip, user_agent: user_agent, operation: operation, status: status, **kwargs } # 记录到文件日志 logger.info(f{operation} - {status} - IP: {client_ip} - {kwargs.get(file_name, N/A)}) # 记录到数据库 try: conn sqlite3.connect(deoldify_audit.db) c conn.cursor() c.execute( INSERT INTO audit_logs (timestamp, client_ip, user_agent, operation, file_name, file_size, file_type, request_id, status, processing_time, error_message, additional_info) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?) , ( log_entry[timestamp], log_entry[client_ip], log_entry[user_agent], log_entry[operation], log_entry.get(file_name), log_entry.get(file_size), log_entry.get(file_type), log_entry.get(request_id), log_entry[status], log_entry.get(processing_time), log_entry.get(error_message), json.dumps(log_entry.get(additional_info, {})) )) conn.commit() conn.close() except Exception as e: logger.error(fFailed to save audit log to database: {e}) # 装饰器自动记录处理时间 def audit_log(operation_name): def decorator(func): def wrapper(*args, **kwargs): start_time time.time() request_id freq_{int(start_time * 1000)}_{os.urandom(4).hex()} try: # 记录开始日志 log_audit_event( operationoperation_name, statusstarted, request_idrequest_id, additional_info{args: str(args), kwargs: str(kwargs)} ) # 执行原函数 result func(*args, **kwargs) # 记录成功日志 processing_time time.time() - start_time log_audit_event( operationoperation_name, statussuccess, request_idrequest_id, processing_timeprocessing_time ) return result except Exception as e: # 记录失败日志 processing_time time.time() - start_time log_audit_event( operationoperation_name, statusfailed, request_idrequest_id, processing_timeprocessing_time, error_messagestr(e) ) raise e return wrapper return decorator # DeOldify处理端点添加审计日志 app.route(/colorize, methods[POST]) audit_log(image_upload_and_process) def colorize_image(): 处理图片上色请求 # 检查是否有文件 if image not in request.files: return jsonify({success: False, error: No image file provided}), 400 file request.files[image] # 记录文件信息 file_info { file_name: file.filename, file_size: len(file.read()), file_type: file.content_type } # 重置文件指针 file.seek(0) # 这里添加额外的文件信息到日志上下文 # 在实际实现中可能需要修改装饰器来传递这些信息 # 模拟处理过程 time.sleep(2) # 模拟处理时间 # 返回处理结果 return jsonify({ success: True, message: Image processed successfully, request_id: 模拟请求ID }) # 下载端点添加审计日志 app.route(/download/request_id, methods[GET]) audit_log(result_download) def download_result(request_id): 下载处理结果 # 模拟下载过程 time.sleep(0.5) return jsonify({ success: True, message: fDownloaded result for request {request_id} }) # 健康检查端点 app.route(/health, methods[GET]) def health_check(): 健康检查 return jsonify({ status: healthy, service: deoldify-colorization, timestamp: datetime.now().isoformat() }) # 审计日志查询端点 app.route(/audit/logs, methods[GET]) def get_audit_logs(): 查询审计日志 # 获取查询参数 operation request.args.get(operation) status request.args.get(status) start_time request.args.get(start_time) end_time request.args.get(end_time) limit int(request.args.get(limit, 100)) # 构建查询 query SELECT * FROM audit_logs WHERE 11 params [] if operation: query AND operation ? params.append(operation) if status: query AND status ? params.append(status) if start_time: query AND timestamp ? params.append(start_time) if end_time: query AND timestamp ? params.append(end_time) query ORDER BY timestamp DESC LIMIT ? params.append(limit) # 执行查询 conn sqlite3.connect(deoldify_audit.db) conn.row_factory sqlite3.Row # 返回字典格式 c conn.cursor() c.execute(query, params) rows c.fetchall() # 转换为字典列表 logs [] for row in rows: log_dict dict(row) if log_dict[additional_info]: log_dict[additional_info] json.loads(log_dict[additional_info]) logs.append(log_dict) conn.close() return jsonify({ success: True, count: len(logs), logs: logs }) # 服务统计端点 app.route(/audit/stats, methods[GET]) def get_audit_stats(): 获取服务统计信息 conn sqlite3.connect(deoldify_audit.db) c conn.cursor() # 总请求数 c.execute(SELECT COUNT(*) FROM audit_logs) total_requests c.fetchone()[0] # 成功/失败统计 c.execute(SELECT status, COUNT(*) FROM audit_logs GROUP BY status) status_stats dict(c.fetchall()) # 操作类型统计 c.execute(SELECT operation, COUNT(*) FROM audit_logs GROUP BY operation) operation_stats dict(c.fetchall()) # 今日请求数 today datetime.now().date().isoformat() c.execute(SELECT COUNT(*) FROM audit_logs WHERE timestamp LIKE ?, (f{today}%,)) today_requests c.fetchone()[0] # 平均处理时间仅限成功请求 c.execute(SELECT AVG(processing_time) FROM audit_logs WHERE statussuccess AND processing_time IS NOT NULL) avg_processing_time c.fetchone()[0] or 0 conn.close() return jsonify({ success: True, stats: { total_requests: total_requests, today_requests: today_requests, status_distribution: status_stats, operation_distribution: operation_stats, average_processing_time_seconds: round(avg_processing_time, 2) } }) if __name__ __main__: # 初始化数据库 init_db() # 记录服务启动 logger.info(DeOldify service starting with audit logging enabled) app.run(host0.0.0.0, port7860, debugTrue)3.2 日志查询工具为了方便查看和分析日志我们可以创建一个命令行工具#!/usr/bin/env python3 DeOldify审计日志查询工具 import sqlite3 import json import argparse from datetime import datetime, timedelta import sys def query_logs(db_path, operationNone, statusNone, start_dateNone, end_dateNone, limit50): 查询审计日志 conn sqlite3.connect(db_path) conn.row_factory sqlite3.Row c conn.cursor() # 构建查询条件 conditions [] params [] if operation: conditions.append(operation ?) params.append(operation) if status: conditions.append(status ?) params.append(status) if start_date: conditions.append(timestamp ?) params.append(start_date) if end_date: conditions.append(timestamp ?) params.append(end_date) # 构建SQL where_clause AND .join(conditions) if conditions else 11 query f SELECT * FROM audit_logs WHERE {where_clause} ORDER BY timestamp DESC LIMIT ? params.append(limit) # 执行查询 c.execute(query, params) rows c.fetchall() # 输出结果 print(f\n{*80}) print(f审计日志查询结果 (共 {len(rows)} 条记录)) print(f{*80}\n) for row in rows: log dict(row) print(f时间: {log[timestamp]}) print(f操作: {log[operation]} - {log[status]}) print(f客户端: {log[client_ip]}) if log[file_name]: print(f文件: {log[file_name]} ({log[file_size]} bytes)) if log[processing_time]: print(f处理时间: {log[processing_time]:.2f}秒) if log[error_message]: print(f错误: {log[error_message]}) print(f{-*40}) conn.close() def show_stats(db_path, days7): 显示统计信息 conn sqlite3.connect(db_path) c conn.cursor() # 计算日期范围 end_date datetime.now() start_date end_date - timedelta(daysdays) print(f\n{*80}) print(fDeOldify服务统计 ({start_date.date()} 至 {end_date.date()})) print(f{*80}\n) # 总请求数 c.execute( SELECT COUNT(*) FROM audit_logs WHERE timestamp ? AND timestamp ? , (start_date.isoformat(), end_date.isoformat())) total c.fetchone()[0] print(f 总请求数: {total}) # 按操作类型统计 print(f\n 操作类型分布:) c.execute( SELECT operation, COUNT(*) as count FROM audit_logs WHERE timestamp ? AND timestamp ? GROUP BY operation ORDER BY count DESC , (start_date.isoformat(), end_date.isoformat())) for op, count in c.fetchall(): percentage (count / total * 100) if total 0 else 0 print(f {op}: {count} 次 ({percentage:.1f}%)) # 按状态统计 print(f\n 成功/失败统计:) c.execute( SELECT status, COUNT(*) as count FROM audit_logs WHERE timestamp ? AND timestamp ? GROUP BY status ORDER BY count DESC , (start_date.isoformat(), end_date.isoformat())) for status, count in c.fetchall(): percentage (count / total * 100) if total 0 else 0 print(f {status}: {count} 次 ({percentage:.1f}%)) # 平均处理时间 c.execute( SELECT AVG(processing_time) FROM audit_logs WHERE statussuccess AND processing_time IS NOT NULL AND timestamp ? AND timestamp ? , (start_date.isoformat(), end_date.isoformat())) avg_time c.fetchone()[0] or 0 print(f\n⏱ 平均处理时间: {avg_time:.2f}秒) # 最活跃的客户端 print(f\n 最活跃客户端 (Top 5):) c.execute( SELECT client_ip, COUNT(*) as count FROM audit_logs WHERE timestamp ? AND timestamp ? AND client_ip ! N/A GROUP BY client_ip ORDER BY count DESC LIMIT 5 , (start_date.isoformat(), end_date.isoformat())) for ip, count in c.fetchall(): print(f {ip}: {count} 次请求) # 每日请求趋势 print(f\n 每日请求趋势:) c.execute( SELECT DATE(timestamp) as day, COUNT(*) as count FROM audit_logs WHERE timestamp ? AND timestamp ? GROUP BY DATE(timestamp) ORDER BY day , (start_date.isoformat(), end_date.isoformat())) for day, count in c.fetchall(): print(f {day}: {count} 次请求) conn.close() def export_logs(db_path, output_file, formatjson): 导出日志 conn sqlite3.connect(db_path) conn.row_factory sqlite3.Row c conn.cursor() c.execute(SELECT * FROM audit_logs ORDER BY timestamp DESC) rows c.fetchall() logs [] for row in rows: log dict(row) if log[additional_info]: log[additional_info] json.loads(log[additional_info]) logs.append(log) if format json: with open(output_file, w, encodingutf-8) as f: json.dump(logs, f, ensure_asciiFalse, indent2) print(f 日志已导出到 {output_file} (JSON格式)) elif format csv: import csv with open(output_file, w, newline, encodingutf-8) as f: if logs: writer csv.DictWriter(f, fieldnameslogs[0].keys()) writer.writeheader() writer.writerows(logs) print(f 日志已导出到 {output_file} (CSV格式)) conn.close() def main(): 主函数 parser argparse.ArgumentParser(descriptionDeOldify审计日志查询工具) parser.add_argument(--db, defaultdeoldify_audit.db, help数据库文件路径 (默认: deoldify_audit.db)) subparsers parser.add_subparsers(destcommand, help子命令) # 查询命令 query_parser subparsers.add_parser(query, help查询日志) query_parser.add_argument(--operation, help操作类型) query_parser.add_argument(--status, help状态) query_parser.add_argument(--start-date, help开始日期 (YYYY-MM-DD)) query_parser.add_argument(--end-date, help结束日期 (YYYY-MM-DD)) query_parser.add_argument(--limit, typeint, default50, help返回条数) # 统计命令 stats_parser subparsers.add_parser(stats, help查看统计) stats_parser.add_argument(--days, typeint, default7, help统计天数 (默认: 7天)) # 导出命令 export_parser subparsers.add_parser(export, help导出日志) export_parser.add_argument(--output, requiredTrue, help输出文件) export_parser.add_argument(--format, choices[json, csv], defaultjson, help导出格式) args parser.parse_args() if args.command query: query_logs( args.db, operationargs.operation, statusargs.status, start_dateargs.start_date, end_dateargs.end_date, limitargs.limit ) elif args.command stats: show_stats(args.db, daysargs.days) elif args.command export: export_logs(args.db, args.output, formatargs.format) else: parser.print_help() if __name__ __main__: main()3.3 日志轮转配置为了防止日志文件无限增长我们需要配置日志轮转import logging from logging.handlers import RotatingFileHandler import os def setup_logging(): 配置日志系统 # 创建日志目录 log_dir logs os.makedirs(log_dir, exist_okTrue) # 审计日志 - 按大小轮转 audit_handler RotatingFileHandler( filenameos.path.join(log_dir, audit.log), maxBytes10*1024*1024, # 10MB backupCount10, # 保留10个备份 encodingutf-8 ) audit_handler.setFormatter(logging.Formatter( %(asctime)s - %(name)s - %(levelname)s - %(message)s )) # 错误日志 - 按时间轮转 error_handler RotatingFileHandler( filenameos.path.join(log_dir, error.log), maxBytes5*1024*1024, # 5MB backupCount5, # 保留5个备份 encodingutf-8 ) error_handler.setLevel(logging.ERROR) error_handler.setFormatter(logging.Formatter( %(asctime)s - %(name)s - %(levelname)s - %(pathname)s:%(lineno)d - %(message)s )) # 控制台输出 console_handler logging.StreamHandler() console_handler.setFormatter(logging.Formatter( %(asctime)s - %(levelname)s - %(message)s )) # 配置根日志 logging.basicConfig( levellogging.INFO, handlers[audit_handler, error_handler, console_handler] )4. 审计日志的实际应用4.1 监控面板我们可以创建一个简单的Web监控面板来实时查看服务状态from flask import Flask, render_template import sqlite3 from datetime import datetime, timedelta import json app Flask(__name__) app.route(/dashboard) def dashboard(): 监控面板 conn sqlite3.connect(deoldify_audit.db) c conn.cursor() # 获取今日数据 today datetime.now().date().isoformat() # 今日请求数 c.execute( SELECT COUNT(*) FROM audit_logs WHERE timestamp LIKE ? , (f{today}%,)) today_requests c.fetchone()[0] # 今日成功/失败数 c.execute( SELECT status, COUNT(*) FROM audit_logs WHERE timestamp LIKE ? GROUP BY status , (f{today}%,)) today_status dict(c.fetchall()) # 最近1小时请求趋势 one_hour_ago (datetime.now() - timedelta(hours1)).isoformat() c.execute( SELECT strftime(%Y-%m-%d %H:%M, timestamp) as minute, COUNT(*) as count FROM audit_logs WHERE timestamp ? GROUP BY strftime(%Y-%m-%d %H:%M, timestamp) ORDER BY minute , (one_hour_ago,)) hourly_trend [] for minute, count in c.fetchall(): hourly_trend.append({ time: minute, count: count }) # 最近10条日志 c.execute( SELECT timestamp, operation, status, client_ip, file_name FROM audit_logs ORDER BY timestamp DESC LIMIT 10 ) recent_logs [] for row in c.fetchall(): recent_logs.append({ timestamp: row[0], operation: row[1], status: row[2], client_ip: row[3], file_name: row[4] or N/A }) conn.close() return render_template(dashboard.html, today_requeststoday_requests, today_statustoday_status, hourly_trendjson.dumps(hourly_trend), recent_logsrecent_logs) if __name__ __main__: app.run(port8080)对应的HTML模板!DOCTYPE html html head titleDeOldify服务监控面板/title script srchttps://cdn.jsdelivr.net/npm/chart.js/script style body { font-family: Arial, sans-serif; margin: 20px; } .dashboard { display: grid; grid-template-columns: 1fr 1fr; gap: 20px; } .card { background: #f5f5f5; padding: 20px; border-radius: 8px; } .stats { display: grid; grid-template-columns: repeat(3, 1fr); gap: 10px; } .stat-item { text-align: center; padding: 10px; } .success { color: green; } .failed { color: red; } table { width: 100%; border-collapse: collapse; } th, td { padding: 8px; text-align: left; border-bottom: 1px solid #ddd; } /style /head body h1 DeOldify服务监控面板/h1 div classdashboard div classcard h2 今日概览/h2 div classstats div classstat-item div classstat-value{{ today_requests }}/div div classstat-label总请求数/div /div div classstat-item success div classstat-value{{ today_status.get(success, 0) }}/div div classstat-label成功/div /div div classstat-item failed div classstat-value{{ today_status.get(failed, 0) }}/div div classstat-label失败/div /div /div /div div classcard h2 最近1小时请求趋势/h2 canvas idtrendChart width400 height200/canvas /div div classcard stylegrid-column: span 2; h2 最近操作记录/h2 table thead tr th时间/th th操作/th th状态/th th客户端IP/th th文件/th /tr /thead tbody {% for log in recent_logs %} tr td{{ log.timestamp }}/td td{{ log.operation }}/td td class{{ success if log.status success else failed }} {{ log.status }} /td td{{ log.client_ip }}/td td{{ log.file_name }}/td /tr {% endfor %} /tbody /table /div /div script // 渲染趋势图表 const trendData {{ hourly_trend|safe }}; const ctx document.getElementById(trendChart).getContext(2d); new Chart(ctx, { type: line, data: { labels: trendData.map(item item.time.split( )[1]), datasets: [{ label: 请求数, data: trendData.map(item item.count), borderColor: rgb(75, 192, 192), tension: 0.1 }] }, options: { responsive: true, plugins: { legend: { display: false } } } }); /script /body /html4.2 自动化报告我们可以设置定时任务每天自动发送服务报告import smtplib from email.mime.text import MIMEText from email.mime.multipart import MIMEMultipart import sqlite3 from datetime import datetime, timedelta def generate_daily_report(): 生成每日报告 conn sqlite3.connect(deoldify_audit.db) c conn.cursor() # 获取昨日日期 yesterday (datetime.now() - timedelta(days1)).date().isoformat() # 昨日统计 c.execute( SELECT COUNT(*) as total, SUM(CASE WHEN statussuccess THEN 1 ELSE 0 END) as success, SUM(CASE WHEN statusfailed THEN 1 ELSE 0 END) as failed, AVG(CASE WHEN statussuccess THEN processing_time ELSE NULL END) as avg_time FROM audit_logs WHERE timestamp LIKE ? , (f{yesterday}%,)) total, success, failed, avg_time c.fetchone() success_rate (success / total * 100) if total 0 else 0 # 最活跃客户端 c.execute( SELECT client_ip, COUNT(*) as count FROM audit_logs WHERE timestamp LIKE ? AND client_ip ! N/A GROUP BY client_ip ORDER BY count DESC LIMIT 3 , (f{yesterday}%,)) top_clients c.fetchall() # 最常见的错误 c.execute( SELECT error_message, COUNT(*) as count FROM audit_logs WHERE timestamp LIKE ? AND statusfailed AND error_message IS NOT NULL GROUP BY error_message ORDER BY count DESC LIMIT 5 , (f{yesterday}%,)) common_errors c.fetchall() conn.close() # 生成报告内容 report f DeOldify服务日报 ({yesterday}) 总体统计: --------- • 总请求数: {total} • 成功: {success} ({success_rate:.1f}%) • 失败: {failed} • 平均处理时间: {avg_time:.2f}秒 最活跃客户端 (Top 3): -------------------- for ip, count in top_clients: report f• {ip}: {count} 次请求\n if common_errors: report f 最常见的错误: ------------- for error, count in common_errors: report f• {error}: {count} 次\n report f 报告生成时间: {datetime.now().strftime(%Y-%m-%d %H:%M:%S)} return report def send_email_report(receiver_email, report_content): 发送邮件报告 # 邮件配置需要根据实际情况修改 sender_email deoldifyexample.com password your_password smtp_server smtp.example.com smtp_port 587 # 创建邮件 msg MIMEMultipart() msg[From] sender_email msg[To] receiver_email msg[Subject] fDeOldify服务日报 - {datetime.now().date()} # 添加正文 msg.attach(MIMEText(report_content, plain)) try: # 发送邮件 server smtplib.SMTP(smtp_server, smtp_port) server.starttls() server.login(sender_email, password) server.send_message(msg) server.quit() print(f 日报已发送到 {receiver_email}) except Exception as e: print(f 发送邮件失败: {e}) # 定时发送报告可以使用cron或systemd timer if __name__ __main__: report generate_daily_report() print(report) # 发送给管理员 send_email_report(adminexample.com, report)5. 总结通过为DeOldify服务添加审计日志系统我们实现了5.1 核心功能回顾完整的行为记录记录了所有图片上传、处理、下载操作的时间戳和详细信息双重存储机制文件日志用于日常查看数据库用于统计分析灵活的查询工具提供了命令行工具和Web接口来查询和分析日志实时监控面板可以实时查看服务状态和请求趋势自动化报告定期生成服务使用报告帮助了解服务运行状况5.2 实际价值问题排查当用户反馈处理失败时可以通过请求ID快速定位问题使用分析了解服务的使用模式优化资源分配安全监控及时发现异常访问模式服务优化根据处理时间统计优化服务性能合规审计满足数据处理的审计要求5.3 扩展建议如果你需要更强大的审计功能可以考虑集成专业日志系统如ELK StackElasticsearch, Logstash, Kibana添加用户认证记录具体用户的操作而不仅仅是IP地址实现日志加密对敏感信息进行加密存储添加告警机制当出现大量失败请求时自动告警集成到现有监控系统如Prometheus Grafana审计日志不是可有可无的装饰品而是服务可靠运行的重要保障。对于像DeOldify这样的AI服务完整的操作记录不仅能帮助你更好地管理服务还能在出现问题时提供关键的排查线索。获取更多AI镜像想探索更多AI镜像和应用场景访问 CSDN星图镜像广场提供丰富的预置镜像覆盖大模型推理、图像生成、视频生成、模型微调等多个领域支持一键部署。