0

Today I accidentally restored a pg_dumpall file to the database that that dumpall was generated from. This restore has caused a ton of duplicates of certain ID's and such.

This is a complex production database and it hasn't corrupted the entire database but its really affecting the ability to change things on the site that uses it. Now I know I could just cut my losses and reset the database to the original pg_dumpall (it was as recent as last night) but that would mean that a full day's work/posting on the site that uses the database would be lost.

Is there anything I can do to remedy this situation?

4

1 回答 1

1

没有主键，这很难。使用主键，这不可能发生。你能做的最好的事情是，恕我直言：

  1. 重命名所有模式（例如schema-> bad_schema）
  2. 重新导入昨晚的转储（当然不删除数据库）
  3. 为所有新表添加 PK 约束（这需要手动工作！）
  4. for all schemas.tables：添加行 from bad_schema.tableXXXtoschema.tableXXX 如果它们还不存在（这可能需要（非唯一）索引bad_schema.XXX，对应于预期的 FK）
  5. （可选）将序列调整为相应列中的现有最大值（值）。

如果您不了解上述内容（或者工作量太大），请承担您的损失，并安装昨天的备份。您可以先重命名损坏的数据库（或重命名模式），然后再尝试从中提取新记录。

摘要：确实意识到每个（部分）解决方案都以某种方式需要主键。没有PK的表是脆弱的，毫无意义的。

于 2016-06-09T20:32:22.233 回答