瀚高数据库开启加密（sm4）导致pg_resetwal不生效问题

瀚高PG实验室 2025-08-18

153

症状

无法启动数据库，报错ould not locate a valid checkpoint record，尝试使用pg_resetwal修复。修复后无效。相关报错信息如下

pg_ctl 启动报错

[root@cxt pg_wal]# pg_ctl start
等待服务器进程启动 ....2024-01-31 15:31:12.901 CST [258179] LOG:  data encryption performed by sm4  《《《
2024-01-31 15:31:12.903 CST [258179] LOG:  starting HighGo Security Enterprise Edition Database System 4.5.8.2 on x86_64,build on 20240119
2024-01-31 15:31:12.903 CST [258179] LOG:  listening on IPv4 address "0.0.0.0", port 5871
2024-01-31 15:31:12.903 CST [258179] LOG:  listening on IPv6 address "::", port 5871
2024-01-31 15:31:12.904 CST [258179] LOG:  listening on Unix socket "/tmp/.s.PGSQL.5871"
2024-01-31 15:31:12.913 CST [258179] LOG:  redirecting log output to logging collector process
2024-01-31 15:31:12.913 CST [258179] HINT:  Future log output will appear in directory "hgdb_log".
........................................................... 已停止等待
pg_ctl: 服务没有及时启动
[root@cxt pg_wal]#

查看日志错误信息

2024-01-31 15:31:12.915 CST,,,258182,,65b9f740.3f086,1,,2024-01-31 15:31:12 CST,,0,LOG,00000,"database system was interrupted; last known up at 2024-01-31 15:26:03 CST",,,,,,,,,""
2024-01-31 15:31:13.617 CST,,,258182,,65b9f740.3f086,2,,2024-01-31 15:31:12 CST,,0,LOG,00000,"invalid primary checkpoint record",,,,,,,,,""
2024-01-31 15:31:13.617 CST,,,258182,,65b9f740.3f086,3,,2024-01-31 15:31:12 CST,,0,PANIC,XX000,"could not locate a valid checkpoint record",,,,,,,,,""
2024-01-31 15:31:13.794 CST,,,258179,,65b9f740.3f083,3,,2024-01-31 15:31:12 CST,,0,LOG,00000,"startup process (PID 258182) was terminated by signal 6: Aborted",,,,,,,,,""
2024-01-31 15:31:13.794 CST,,,258179,,65b9f740.3f083,4,,2024-01-31 15:31:12 CST,,0,LOG,00000,"aborting startup due to startup process failure",,,,,,,,,""

数据库报错，could not locate a valid checkpoint record。说明控制文件或wal日志有损坏。此时需要使用备份恢复。如果没有备份，可尝试pg_resetwal。其会清空旧的wal，产生新的wal，并更新控制文件信息。

pg_resetwal修复，启动后依然报错。

pg_resetwal -D /opt/highgo/hgdb-see-4.5.8/data -f
pg_ctl start  #无法启动

触发条件

检查数据库启用了sm4。 pg_controldata |grep encryption

解决方案

应用数据库补丁，如不能使用补丁。则考虑备份恢复。

hgdb458以补丁的形式提供。补丁包hgdb-see-4.5.8-patch-0330-20240119.x86_64.rpm
hgdb457以整包的形式提供。安装包hgdb-see-4.5.7-da8c022.x86_64.rpm
hgdb456以整包的形式提供。安装包hgdb-see-4.5.6-4c56395.x86_64.rpm

下面以hgdb458应用补丁为例

1. 上传和安装补丁

rpm -ivh hgdb-see-4.5.8-patch-0330-20240119.x86_64.rpm

补丁安装路径：$HGDB_HOME/etc/patch-20240119，此路径下有README.txt

2. 应用补丁并启动数据库

#关闭数据库
pg_ctl stop
#应用补丁
cd $HGDB_HOME/etc/patch-20240119
./upgrade.sh patch-20240119

3. 重新resetwal

pg_resetwal -D /opt/hgdata -f

4. 启动数据库，数据库可正常启动，resetwal生效

pg_ctl start
#数据库版本变为4582
highgo=# select version();
                           version                          
-------------------------------------------------------------
 瀚高安全版数据库系统V4.5.8.2, x86_64平台, 创建日期:20240119
(1 行记录)

文章转载自瀚高PG实验室，如果涉嫌侵权，请发送邮件至：contact@modb.pro进行举报，并提供相关证据，一经查实，墨天轮将立刻删除相关内容。

瀚高数据库开启加密（sm4）导致pg_resetwal不生效问题

评论