Restoring attachments from cloud to server

[David Martines]

We are migrating some projects from JIRA cloud to an older (7.3.6) on-prem server instance.   We originally created a backup file from cloud and tried to import projects from it into server, but had all sorts of incompatibility problems.  We determined just importing the issues via external system import / CSV would be fine.   

After importing the issues from CSV, next task is to get the attachments.  I found the attachments in the original backup zip and copied them to the attachments dir on the server.  However, even though the file names and paths are correct (files are named by attachment ID), they will not load because they are not actually images (png) but are html snippets (of what looks the jira cloud login).  I figured this was due to the backup being possibly created for "cloud" and not server, so we created a new backup, ensuing "for server" option was used. 

This time the attachment files are actual image data - but the names are all wrong.. the files do not correspond to the attachment IDs, thus won't load.  

Any advice on how to proceed with restoring attachments?
Thanks

[Normann P_ Nielsen _Netic_]

I dont think You can save the situation, and get stuff linked - within reasonable time/effort.

You dont write what "but had all sorts of incompatibility problems.  " - but I guess its Oracle and possibly MS SQL (I spend 1 year trying to migrate from Cloud to Oracle, and even Atlassian Support used month on it - we gave up and used Postgress)

Exporting from Cloud is (as far as I know from Postgress) - and should be relatively easily imported into Postgres and possibly MySQL.

You might need to run the xml file through the XML Cleaner - https://confluence.atlassian.com/jira/removing-invalid-characters-from-xml-backups-12079.html

I needed that 2 weeks ago - even from cloud 2 cloud with a cloud backup... (go figure).

BR,

Normann

[David Martines]

Problems with direct project import from backup zip, were issue types, type schemes, custom fields, workflows, etc. all mismatched between source and target systems.  Hence external system import of issues only, from CSV.

Anyway, problem solved.  The reason the import process created all the attachment files as html snippets of the Atlassian login screen, was that the source cloud jira instance is not public, so the target system was not able to download the attachments from their URLs in the CSV.   Instead of the file, it just saved the html of the login screen.

What I should have done was copy all attachment files from backup zip to a staging folder, and then change all URLs in the CSV to use the file:/// protocol to point to this staging area.