QOS_MESSAGE temporary queue for the dashboard_engine stops draining (being processed by dashboard_engine). It gets very large and constantly we have to disconnect it in the hub. It seems that the dashboard_engine just stops processing messages.
This is a workaround to disable the dashboard_engine temp queues, and get the information in a more efficient method. This describes how to avoid any adverse performance impact on the hub from unnecessary dashboard_engine temp queues.
Restarting the dashboard_engine temporarily relieves the issue but the temp queue(s) will start to build again. The basic problem is "too much QoS". This is a workaround to disable the temp queues, and get the information in a more efficient method.
In large environments using nis_server, it is possible for the volume of QoS measurements to overwhelm the hub, causing constant QoS message queues to form. If this occurs, the hub's performance may not be optimal. In addition, the summary lists in the Dynamic Views portlet in UMP may not display up-to-date QoS data.
As an option, you can disable dashboard_engine 'QoS listening' to reduce the hub's workload.
Note: Disabling QoS listening results in a number of empty columns in the Summary views in Dynamic Views. The dashboard_engine's functionality is not affected in other UMP portlets, such as Alarm Console, Custom Dashboards, and Dynamic View Dashboards. This procedure includes an optional step to fix the empty columns that result in the Summary views.
To disable QoS listening, follow these steps:
dashboard_engine (same as dap) is deprecated and no longer needed for UMP to work.
If this issue is affecting any version of UIM > 7.1 (8.x, 9.x) Dashboard_engine probe can be disable/delete and this issue will no longer occur.