PostgreSQL - 如何恢复非常大的数据库

Question

当我尝试在我的开发数据库节点(这个节点的 RAM、CPU... 比我的生产服务器少)上恢复一个大型数据库(自定义格式几乎 32Go)时我遇到了一个小问题。

我的数据库转储是使用类似于以下的命令生成的:

pg_dump -F custom -b myDB -Z 9 > /backup/myDB-`date +%y%m%d`.pg91

当我恢复它时，我使用了以下命令:

pg_restore -F custom -j 5 -d myDB /backup/myDB-20130331.pg91

但在这里，每次恢复命令失败时都会出现如下错误:

pg_restore: [archiver (db)] error returned by PQputCopyData: server closed the connection unexpectedly
This probably means the server terminated abnormally
    before or while processing the request.
pg_restore: [archiver] worker process failed: exit code 1
pg_restore: [archiver (db)] error returned by PQputCopyData: server closed the connection unexpectedly
This probably means the server terminated abnormally
    before or while processing the request.
pg_restore: [archiver (db)] error returned by PQputCopyData: server closed the connection unexpectedly
This probably means the server terminated abnormally
    before or while processing the request.
pg_restore: [archiver (db)] error returned by PQputCopyData: server closed the connection unexpectedly
This probably means the server terminated abnormally
    before or while processing the request.

当我检查我的 postgresql 日志时，我可以读到:

   HINT:  In a moment you should be able to reconnect to the database and repeat your command.
   LOG:  all server processes terminated; reinitializing
   LOG:  database system was interrupted; last known up at 2013-04-02 11:41:48 UTC
   LOG:  database system was not properly shut down; automatic recovery in progress
   LOG:  redo starts at 86/26F302B0
   LOG:  unexpected pageaddr 85/E3F52000 in log file 134, segment 38, offset 16064512
   LOG:  redo done at 86/26F51FC0
   LOG:  last completed transaction was at log time 2013-04-02 11:50:47.663599+00
   LOG:  database system is ready to accept connections
   LOG:  autovacuum launcher started

很奇怪，我的 postgresql 服务器仅仅因为我的恢复就单独“重启”了。我尽量减少作业数量(-j 5 选项)但仍然遇到同样的问题。但是在具有更好规范的节点上，我可以毫无问题地恢复该数据库。我不确定，但也许我的索引更新(其中一个确实太大)可能是理解这个问题的线索？

所以我有一些问题:是否有更好的方法来恢复非常大的数据库？我在我的 pg_restore 命令中遗漏了什么吗？可能是我的开发服务器设置太低了吗？

任何线索将不胜感激。提前致谢。

环境:PostgreSQL 9.1(通过 Debian 软件包安装)

Answer 1

对于这种大型工作，建议禁用 autovacuum(在 postgresql.conf 中将其设置为 off)在修复过程中。

看来它终于对我有用了。