Proper approach to manipulate Confluence Database

[MichaelMoe]

I'm working on a project that imports contents into Confluence. Since Confluence does not have masquerade feature(well, that's what JIRA ticket from couple of years ago suggested and there's still no update on the ticket.) , I manipulate Confluence Database directly by issuing SQL statements to restore original content creation date, original content creator, likes, and last modified date. To see the changes, I need to flush Confluence Cache and rebuild index. It is not an issue at this point as I'm working on test site. But, soon I'll be working on production site and I'm concern that this may corrupt the database. What should I do to prevent corrupting database?

[Nic Brough -Adaptavist- · Accepted solution]

• Test your SQL to death on a test machine
• Backup Confluence before running it on production
• Have Confluence offline when you run the SQL
• Restart and trigger a re-index

Your main problems are the caches and index, taking Confluence offline while you run SQL is the only way you can avoid damage to the database.  Ideally, avoid SQL like the plague, it's a last resort for working with Confluence (or JIRA)

[MichaelMoe]

Good points. Thanks!

[Richard Atkins]

The relevant CONF issue for impersonating users: https://jira.atlassian.com/browse/CONF-6786