Uploaded image for project: 'Confluence Server'
  1. Confluence Server
  2. CONFSERVER-46178

Synchrony Status Cache doesn't work in cluster


    • Symptom Severity:
    • Sprint:
      Paddington, Queenscliff
    • QA Demo Status:
      Not Needed
    • QA Kickoff Status:
      Not Needed


      The SynchronyStatusCheckJob runs every 30 seconds but doesn't actually save anything to the cache. This is most-likely a bug in VCache to do with the Hazelcast cache. It works fine in non-DC instances.

      Because of this, Synchrony might be down but the big red banner will never display, since if there is nothing in the cache, we assume Synchrony is just up and everything is fine. That should only happen the first time the job runs. Every other time it will be populated with Synchrony's actual status.

      We should fix the VCache issue with Hazelcast. We shouldn't use a supplier in the get for the cache miss, because it will make all page loads slow since it is used by the SynchronyDownBannerDataProvider which determines if the banner should be displayed or not. That runs on every page, and we can't be waiting for a rest request there, even if it's only for Sysadmins. Additionally, it accesses the cache multiple times in the job itself.




            • Votes:
              0 Vote for this issue
              2 Start watching this issue


              • Created:
                Last commented:
                1 year, 10 weeks, 5 days ago