For those QlikView administrators that wish to gain additional information that may not be available from the default logs, there are Windows Performance Monitor counters available for some QlikView Services that may provide the information needed.
QlikView Performance Counters:
- QlikView Management Service:
- QvaEngine.Execute call duration [Duration of an action caused by a click in the QlikView Management console and other background tasks not triggered by user interaction]
- TaskCacheManager.TaskStatusFetchThread.GetTaskStatusesFromQDS call duration [Duration of a call to the QlikView Distribution Service to return task status]
- TaskCacheManager.TaskStatusFetchThread.GetTaskStatusesFromQDS call return count [How many task statuses the QDS has responded with]
- QlikView Distribution Service:
- # of already started tasks [If a task is triggered but is ignored due to to it already running, this value increases by 1]
- # of cluster nodes [How many nodes belong to the cluster]
- # of distributions with errors [Distribution tasks with errors and warnings]
- # of dormant tasks [If a task is modified while it is running, the task is considered dormant]
- # of equal load encounters [If all nodes in the cluster have equal load when the trigger happens, this is recorded. EqualLoadSeperator then decides where the task should run.]
- # of failed ClusterFile open [When the QDS fails to open any cluster file this value increases by 1. Example: Lock LoadBalancer.xml with external tool and you should see a value increase by 1 after configured timeout (default 60 seconds).]
- # of failed distributions
- # of failed tasks [The number of tasks that have failed.]
- # of new task result in task result cache [The number of tasks that have finished.]
- # of recovered tasks [When a task is started but the node that "owns" it is not alive.]
- # of running tasks [Number of running tasks on the machine.]
- # of running tasks in task result cache
- # of skipped triggers [In a cluster, if this machine is not least load this value will increase. Signals that the trigger was detected but ignored since it should run elsewhere.]
- # of started distributions
- # of started tasks
- # of tasks [Total number of tasks.]
- # of updated task result in task result cache [In a cluster, number of task results updated compared to the internal cache (other nodes modified).]
- Average cluster file open duration
- Calculated average CPU load
- Current equal load separator [Random number between 1 and 1000 used to pick the QDS if the loads are equal.]
- Current load [In 12.10 we rely on Windows API to give us the current parameters based on which we calculate the load for the machine and in SR5 and up we only used the QVB's data, so if the QVB was not active then the current load could be 0.]
- Current QVB CPU Load [In 12.10 shows systems data, in SR5 and up QVB´s]
- Current QVB engine count
- Current QVB memory usage [In 12.10 shows systems data, in SR5 and up QVB´s]
- Milliseconds to rotate workorder [Not used in 12.10, SR5, 11.20]
- Milliseconds to store workorder XML to disk
- Pulse
- Task result cache average processing time per thread (ms) [Average time it takes for each cache thread to update the task result from files to in-memory. To get a value you need many tasks in system and multiple tasks constantly triggered.]
- Task result cache highest thread processing time (ms) [Highest time of the same as above.]
- Task result cache lowest thread processing time (ms) [Lowest time of the same as above.]
- Task result cache processing threads [Number of threads handling the task result cache.]
- Task result cache total processing time (ms)
