重新启动备用服务器PostgreSQL V13 Window服务器时出现问题

Question

当我尝试启动PostgreSQL备用服务器时，下面提到的日志文件不断地生成并重新生成相同的日志文件。有人知道为什么数据库总是使用相同的日志文件吗？

2021-06-17 13:34:51.906 +03 [9808] LOG:  restored log file "00000001000000820000003C" from archive
2021-06-17 13:34:51.948 +03 [9808] LOG:  invalid record length at 82/3C000688: wanted 24, got 0
2021-06-17 13:34:55.949 +03 [8572] LOG:  started streaming WAL from primary at 82/3C000000 on timeline 1
2021-06-17 13:34:55.957 +03 [8572] FATAL:  could not receive data from WAL stream: ERROR:  requested WAL segment 00000001000000820000003C has already been removed
2021-06-17 13:34:56.904 +03 [9808] LOG:  restored log file "00000001000000820000003C" from archive
2021-06-17 13:34:56.946 +03 [9808] LOG:  invalid record length at 82/3C000688: wanted 24, got 0
2021-06-17 13:35:00.749 +03 [9012] LOG:  started streaming WAL from primary at 82/3C000000 on timeline 1
2021-06-17 13:35:00.756 +03 [9012] FATAL:  could not receive data from WAL stream: ERROR:  requested WAL segment 00000001000000820000003C has already been removed
2021-06-17 13:35:01.913 +03 [9808] LOG:  restored log file "00000001000000820000003C" from archive
2021-06-17 13:35:01.973 +03 [9808] LOG:  invalid record length at 82/3C000688: wanted 24, got 0
2021-06-17 13:35:05.399 +03 [8660] LOG:  started streaming WAL from primary at 82/3C000000 on timeline 1
2021-06-17 13:35:05.420 +03 [8660] FATAL:  could not receive data from WAL stream: ERROR:  requested WAL segment 00000001000000820000003C has already been removed
2021-06-17 13:35:06.925 +03 [9808] LOG:  restored log file "00000001000000820000003C" from archive
2021-06-17 13:35:06.973 +03 [9808] LOG:  invalid record length at 82/3C000688: wanted 24, got 0
2021-06-17 13:35:10.199 +03 [8496] LOG:  started streaming WAL from primary at 82/3C000000 on timeline 1
2021-06-17 13:35:10.212 +03 [8496] FATAL:  could not receive data from WAL stream: ERROR:  requested WAL segment 00000001000000820000003C has already been removed
2021-06-17 13:35:11.916 +03 [9808] LOG:  restored log file "00000001000000820000003C" from archive
2021-06-17 13:35:11.971 +03 [9808] LOG:  invalid record length at 82/3C000688: wanted 24, got 0
2021-06-17 13:35:15.320 +03 [7672] LOG:  started streaming WAL from primary at 82/3C000000 on timeline 1
2021-06-17 13:35:15.335 +03 [7672] FATAL:  could not receive data from WAL stream: ERROR:  requested WAL segment 00000001000000820000003C has already been removed
2021-06-17 13:35:16.915 +03 [9808] LOG:  restored log file "00000001000000820000003C" from archive
2021-06-17 13:35:16.962 +03 [9808] LOG:  invalid record length at 82/3C000688: wanted 24, got 0

Answer 1

invalid record length消息本身是无害的，它表明没有更多的WAL可被发现。

我可以想到两个原因：

• 宕机时间足够长，以至于主服务器实际上不再有相关信息(在pg_wal中查找以000000820000003C结尾的文件)

在这种情况下，请重建备用服务器。

• 时间线已切换，因为主服务器已升级，但备用服务器没有recovery_target_timeline = latest，因此它不遵循时间线切换

在这种情况下，请更改设置并重新启动。