Symptoms

Confluence throws the following error on a single-node deployment shortly after startup:

Fatal error in Confluence cluster: Database is being updated by an instance which is not part of the current cluster. You should check network connections between cluster nodes, especially multicast traffic.

Cause

Two instances of single-node Confluence (unclustered) are connecting to the database.
There are several known causes:

1. Confluence has not been shut down completely and the java process is still running (and connected to the database).
2. Two instances point to the same confluence-home or database and are running simultaneously.
3. Confluence is deployed in the webapps folder and is reloading.
4. A test server has been deployed that is connecting to the same database.
5. A Windows Service is starting automatically, perhaps left over from a different installation.
6. The parameter reloadable=true is set in the war file XML configuration.

Resolution

If the problem is reoccurring,

1. If Confluence is running on a different server (eg a test server), stop the process there. Make sure the confluence.cfg.xml on the other server does not point to your production database.
2. If you are using a standalone deployment:
   1. Check that Confluence is not deployed in the <confluence-install>/webapps folder. This folder should be empty.
3. If you are using a WAR deployment
   1. Check the confluence.xml file for reloadable=true
   2. Check that Confluence is not deployed multiple times in the <confluence-install>/webapps folder (eg duplicated files in this folder)

If it simply did not shut down properly,

In Linux:

1. run ps -a. Identify the java process for Confluence. Note the process ID.
2. run kill -9 <pid>.

In Windows:

1. From the close program box, kill the java.exe process.