Summary

A misconfiguration of the CDN or a CDN service outage may cause the application's UI to become inaccessible. When this happens, we must run a REST command to remove the data from Confluence database and allow the app to work correctly again.

As described in our CDN configuration guide, we can run this REST command if we are unable to remove the CDN from the Confluence UI:

curl -v -u <admin username>:<admin password> -X DELETE http://<your-base-url>/rest/static-asset-caching/configuration

Environment

Confluence Data Center using CDN (Content Delivery Network).

Diagnosis

The following error appears in the logs after running the REST API command to remove the CDN from Confluence.

ERROR [http-nio-8090-exec-3] [common.error.jersey.ThrowableExceptionMapper] toResponse Uncaught exception thrown by REST service: null
 -- url: /rest/static-asset-caching/configuration | traceId: xxxxxxxxxxxxxx | userName: xxxxxxx
java.lang.NullPointerException
	at com.atlassian.plugins.impl.rest.CdnConfigurationResource.delete(CdnConfigurationResource.java:74)

Cause

This issue is caused by missing data in the database, which is expected by the REST command. The Bandana table keeps the CDN records and if some of the data is missing, the error above may occur.

Workaround

We can remove the CDN configuration manually from the database and set the CDN setting to false using SQL commands.

1) Shutdown Confluence on all nodes.

2) Run the following SQL commands on the Confluence database:

delete from BANDANA where bandanakey = 'atlassian.prefix.cdn.url';
update BANDANA set BANDANACONTEXT='_GLOBAL', BANDANAKEY='atlassian.cdn.enabled', BANDANAVALUE='<string>false</string>' where bandanakey = 'atlassian.cdn.enabled';

3) Start one node at a time. Wait for each node to come up before starting the next node.

4) Check on Confluence Administration > General Configuration > Content Delivery Network to verify that it is disabled.