This JAC ticket refers to a bug where the most recently created users would gain additional entries in the people directory with each Confluence restart, until a re-index is performed (at which time the duplicates are removed).

If you encounter a bug where your people directory is displaying duplicates and this isn't fixed by performing a re-index, it's likely that you're affected by CONF-30050.

Please take a look at this Knowledge Base article for more information:
https://confluence.atlassian.com/display/CONFKB/Duplicates+in+the+People+Directory

Also still seeing this in the latest confluence onDemand

Still seeing this in our OnDemand instance, running 5.3-OD-4

I have similar issue with 5.1.2 Build Number 4224. Im using users from jira. I have 4 users in Jira and one user (me) in confluence ware 4 times after reindexing still 2 instances of me.

If the last indexing task flushed before a Confluence restart is user creation, then the restart will cause a duplicate entry for that user to appear in the people directory. It appears that this bug has existed for a while (I tested 5.0 and was able to reproduce the issue), but I believe it was masked by another bug that was recently fixed (the index queue wasn't being flushed on startup).

When the index queue wasn't automatically flushed, the admin user created during installation didn't show up in the people directory until a subsequent task was added to the index queue. In order for duplicates to appear in the people directory, the last action performed before restart had to be user creation. Now that we flush the index queue on startup, any empty instance left untouched until a restart will have a duplicate entry for the admin user.

At this time, we're still investigating possible fixes for this issue. However, rebuilding the search index should remove the duplicate users in the people directory. You will then also need to perform an action that will add a different type of indexing task to the queue before your instance is restarted - for example, creating a page in Confluence. If this doesn't work for you, it's very likely that your situation has a different root cause to what I described above. Please raise a support ticket so that we can investigate further.

I have the same problem in the OnDemand instance we use. We are not using any 3rd party directories. I tried re-indexing via Administration -> Content Indexing but this didn't help. Other suggestions, e.g. removing the index manually, don't work in the OnDemand instance. Obviously I don't have access to the underlying server(s).