Uploaded image for project: 'Confluence Data Center'
  1. Confluence Data Center
  2. CONFSERVER-56303

Improve log, or error message for when Confluence is unable to access itself, when set behind a reverse proxy

    XMLWordPrintable

Details

    • Suggestion
    • Resolution: Unresolved
    • None
    • Macros - Gadgets
    • None
    • 0
    • 3
    • We collect Confluence feedback from various sources, and we evaluate what we've collected when planning our product roadmap. To understand how this piece of feedback will be reviewed, see our Implementation of New Features Policy.

    Description

      When Confluence is set behind a reverse proxy, and incoming traffic is not allowed, using any internal gadgets opens up a HTTP thread, and clicking on the internal gadgets multiple times will eventually exhaust all HTTP threads, causing Confluence to hang, or crash.

      This behavior is similar to CONFSERVER-39393, where when Confluence is unable to access external gadgets, opening up the gadgets causes Confluence to use up it's HTTP threads.

      In this case, the issue is fixed by allowing incoming traffic from Confluence server to the load balancer.

      The issue is that the errors and symptoms of this issue is similar to CONFSERVER-39393, and as such, it's not immediately apparent what the issue is.

      Steps to Reproduce:

      1. Add a proxy configuration for tomcat in server.xml (e.g. proxyName confluence.xyz where confluence.xyz is a load balancer or webserver).
      2. On the confluence.xyz loadbalancer, do not permit incoming traffic from the Confluence application server. Confluence will (mostly) still work, except for gadgets.
      3. On Confluence, click on Insert > Macro. This will cause the page to spin and timeout. Do this 48 times (or whatever your maxThreads is set to).

      Result: 

      1. Confluence is no longer accessible and needs reboot. There are errors in the log hinting at what may have happened but it is not immediately clear.

      Expected:

      1. There should be a clearer log message or perhaps add a health check that would fails. The current situation where there is a time-delayed crash after some use is not too easy to debug.

       

       

      Attachments

        Issue Links

          Activity

            People

              Unassigned Unassigned
              smaran Sattesh M
              Votes:
              3 Vote for this issue
              Watchers:
              3 Start watching this issue

              Dates

                Created:
                Updated: