Confluence 3.5.1 Upgrade Issues - Embedded Crowd for large enterprises

[skeggert]

Have any other organizations with a large user base (15,000+) and number of groups had issues upgrading to Confluence 3.5.1 or Confluence 3.5.x?

What steps or solutions were taken to overcome the obstacles in Atlassian's recent *change in user management (*Embedded Crowd)?

We are looking for solutions that we could easily leverage without using unsupported third-party tools and developing custom scripts.

Any advice on a Confluence 3.5.x or 3.4.x upgrade?

[Colin Goudie · Accepted solution]

Firstly, as you seem to be aware, because of the change in user management, make sure you use your existing osuser.xml/atlassian-user.xml files from your existing installation when first booting Confluence, as it will use these to setup the embedded crowd environment. Most of this comment is in relation to hooking Confluence up to a Crowd installation (Not moving from a Crowd, to embedded Crowd)

Two of our clients had issues in relation to the below issue that is now fixed in Confluence 3.5.5

http://jira.atlassian.com/browse/CONF-22342

One of these clients, due to using MySQL, also has had issues in relation to this issue

https://jira.atlassian.com/browse/CONF-22541

This last issue is being addressed in the upcoming Confluence 3.5.6

We also have run into issues where for some reason, some usernames have spaces appended to them, and this causes issues with the synchronisation. Repairing them, resolved the issue.

So yes, we have had some issues, but I must say, I do like the embedded Crowd functionality. It is great to be able to force a sync at any time to make sure the user repositories in the applications are up to date.

Neat stuff

Like any upgrade, testing this on suitable staging environment is essential

[David at David Simpson Apps]

Also, enabling nested groups for a user base of that size is likely not advisable.