Crowd database support has rarely kept pace with Jira and Confluence which is frustrating because its purpose is to be used with those products. You may not realise this but when running this we co-locate the database with the other products it works with. Even if not in the same instance (like Oracle) on the same server.
I also note that in the current version of Crowd you are removing JNDI data sources, which is currently the only way to make Crowd work with a Highly available database (primary and standby or multiple standby databases). I understand you've opted for a performance parameter on the jdbc definition for postgresql instead which is not able to be used with a standard jdbc HA connection string format. We don't see this as any actual gain in performance due to the nature of the underlying hardware and network. HA ability is far more important even so that we can switch primary and standby databases around during maintenance and not have to reconfigure Crowd each time.
Again, it is baffling that this is not matching the capability of Jira and Confluence. Please align Crowd with those products that it exists to support.
Suggestion
Crowd database support has rarely kept pace with Jira and Confluence which is frustrating because its purpose is to be used with those products. You may not realise this but when running this we co-locate the database with the other products it works with. Even if not in the same instance (like Oracle) on the same server.
I also note that in the current version of Crowd you are removing JNDI data sources, which is currently the only way to make Crowd work with a Highly available database (primary and standby or multiple standby databases). I understand you've opted for a performance parameter on the jdbc definition for postgresql instead which is not able to be used with a standard jdbc HA connection string format. We don't see this as any actual gain in performance due to the nature of the underlying hardware and network. HA ability is far more important even so that we can switch primary and standby databases around during maintenance and not have to reconfigure Crowd each time.
Again, it is baffling that this is not matching the capability of Jira and Confluence. Please align Crowd with those products that it exists to support.