Confluence copy page with internal image lead to absolute url being used

[S. Piller]

Hi Atlassian,

I guess I've found a minor - but yet annoying - bug in Confluence. The use case is the following:

• you have a page with at least 1 uploaded image in it.
• you made at least one copy of the original page
• and then you changed your hostname from http://oldhostname to http://newhostname 

What happens is that the copied page refer to the absolute url of the original image. You can see in your browser debugger requests to http://oldhostname/confluence...

For me, the bug is that Confluence does not recognize its own resources when copying a page, and refer to them with an absolute url. That leads to inconsistencies when for example, you migrate your content from an instance to another, or put a backup of production data in your integration environment. Request are still made to the production host.

Keep using the old DNS alias, pointing to the new physical host, helps only if the copied pages have no access restrictions. It there is, then the image fails to load (you take a redirect), since you need to authenticate under http://oldhostname to access it.

A request like "UPDATE BODYCONTENT SET BODY = REPLACE(BODY, 'http://oldhostname', 'http://newhostname') WHERE BODY LIKE '%http://oldhostname%'" + a reindex seems to fix the issue, but hum... nobody knows if it does not corrupt the content (ie. do we really want to change the old host in every occurences?) and I don't feel comfortable running it...

Is it already known, or should I open a bug report somewhere ?

Thanks!

[Rob Woodgate · Accepted solution]

Hi,

That definitely sounds like a bug.  Search for it at https://jira.atlassian.com and if you can't find it add it as a bug.

[S. Piller]

Oki  ticket is in progress

Thank you!

[Monique Khairuliana]

Hi S. Piller and everyone 

This is the only similar scenario that's currently known to be happening in Confluence https://jira.atlassian.com/browse/CONF-37772