I highly recommend everyone use native file based, and database based backups

The XML backup being not fully supported is kind of crazy, since it is advertised as a feature to perform backups.

That being said, I highly recommend everyone use native file based, and database based backups to backup your Jira instance. It just works 1000x better than XML backup.

We are restoring a 16gb Active Object file and the system is stucked.

Is urgent to restore the functionality also for us.

We are using Jira 8.13.1 Server and Insight latest version with the cmdb

The issue is important to us as well.
We are now trying to migrate to Jira 8.3.0 on Ubuntu 18, Orcale 12, Java 8 and Tomcat 8 and it is still a problem.

We are unable to get the data to load into Jira 8.3.0 without the modifications in this issue which is now a couple months from being 4 years old.

This issue is critical to us, as depending on the moment the import fails we get incoherent data in the restores instance and it is unusable.

Couple of examples (restoring an export from an Oracle 12 instance to a Postgres 9 instance):

• If the import fails when inserting some rows from activeobjects.xml (due to a invalid character encoding), the import is interrupted and many data from active objects is missing, so plugin configuration and data is not imported.
• If the import fails when adding the foreign key constraints (because the export is not coherent due to changes during the export), then the postgres sequence numbers are not restored and Jira starts failing with "DUPLICATE KEY" errors.

Best scenario would be that the restore (import) is done in a transaction if the database backend supports it, so in case of failure the whole transaction is rolled back and the instance is reverted to the pre-import state). As this might require a very high rollback or undo space, it might be an optional feature to include, so when you start the import there is an option to "allow running in a transaction so import is rolled back in case of failure".

My company is using Jira 7.1.7, Oracle 12c, Tomcat 8.0.30 and Java 1.8.0_66.
The Jira upgrade did complete from 6.1.2.
Steps that worked for here:
Install Jira software and setup the user, generate a test license.
Before you import the data on the Administration -> System page. Check your database.
We had sequences for Active Objects (AO in the name). Clean all of the sequences out. There were no tables with these names.
I used this script with Oracle SQL Developer.
BEGIN
FOR i IN (SELECT us.sequence_name FROM USER_SEQUENCES us)
LOOP
EXECUTE IMMEDIATE 'drop sequence '|| i.sequence_name ||'';
END LOOP;
END;
With no existing tables or sequences with AO in the name in the database.
Now import the data file from the other Jira instance.
Login after the reindexing.

I am having the same issues.
Error I am capturing is this:
There was a problem restoring ActiveObjects data for the plugin Atlassian Whitelist API Plugin(com.atlassian.plugins.atlassian-whitelist-api-plugin) #1.18 plugin. Importing table AO_21D670_WHITELIST_RULES failed. Please check the log for details

If you modify the xml and remove the 2 AO_21D670_WHITELIST_RULES sections you get the next plugin if you try to load.
There was a problem restoring ActiveObjects data for the plugin Atlassian Navigation Links Plugin(com.atlassian.plugins.atlassian-nav-links-plugin) #4.2.4 plugin. Importing table AO_38321B_CUSTOM_CONTENT_LINK failed. Please check the log for details.

I actually have a support ticket open for this type of behavior – JSP-250492.
Using a backup xml file, the system restore would fail stating a duplicate entry. Support was never able to recreate it. I made three attempts at the system restore with the same failure every time.

Uhmmm pkirkeby,

We actually need a back up that reproduce the issue mate.

A large activeobjects.xml is not really detailled enough. Could you please provide us with more detail so we can move this one forwards?

Thanks

Cheers,
Os.

The linked bug is Critical but this is Minor - odd.

How does one know if a backup has failed at the moment? Is the XML zip just incomplete in AO tables? I agree that a log message reporting the failure would help, but the truth is that most admins will never see it. What other mechanism could be used to tell admins that there is a problem