Using crowd with JIRA and Confluence on single host server crowd invalidates sessions when you switch between them.

[Dana Frost]

This is for a dev/test configuration where all three services JIRA, Confluence, and Crowd are running on a single host (using different ports). I have crowd working with both applications (JIRA and Confluence) and I can log into each separately. The problem is that as soon as I log into the second application, the first application requires a new login. The only diffrence between the services is the port number

• Crowd: http://host.example.com:8095/
• JIRA: http://host.example.com:8080/
• Confluence: http://host.example.com:8090/

So, each time I log into JIRA, it will invalidate the current Confluence login and ask me to login again. Crowd is setup to use a Delegated/Generic LDAP connection.

[Dana Frost · Accepted solution]

Using diffrent address per application is the way to go. Got this answer from support...

The reason for this behavior is that the cookie is created via the domain. In this case, each service was at http://host.example.com as far as the browser was concerned. Therefore, each new login granted a new cookie which invalidated the previous one. The most common way around this is to either to us a context path (http://host.example.com/confluence) or a mixed domain (http://confluence.host.example.com).

The other way, which you have discovered, is to establish SSO with Crowd. In this situation the cookie (and token) are assigned view Crowd so for each application the token is the same so you are not logged out.

[Timothy]

Have you tried this? Let your apps have a context path. E.g. /jira or /confluence