eG Help

Cassandra is based on a Staged Event Driven Architecture (SEDA). Cassandra separates different tasks into stages connected by a messaging service. Cassandra maintains distinct thread pools for different stages of execution. Each like task is grouped into a stage having a queue and thread pool . Some stages skip the messaging service and queue tasks immediately on a different stage if it exists on the same node. Each of these queues can be backed up if execution at a stage is being over run. This is a common indication of an issue or performance bottleneck. If too many tasks are pending in a thread pool, then administrators should be alerted about those tasks so that they can either further investigate the problem or add additional threads in that thread pool. The CassThrdPoolTest test helps administrators in this process!

This test auto-discovers the thread pools of each stage on the target Cassandra Database node and for each thread pool, reports the number of active tasks and the tasks that were completed per second. In addition, this test reveals the tasks that were pending and the rate at which the tasks were blocked per second. Using this test, administrators can determine the thread pool containing the maximum number of active threads.

Ouputs of the test:One set of results for the target Cassandra Database node being monitored.

Measurement	Description	Measurement Unit	Interpretation
Active_tasks	Indicates the number of active tasks in this thread pool.	Number	Compare the value of this measure across the thread pools to figure out the thread pool that contains the maximum number of threads that are active.
Completed_tasks	Indicates the rate at which tasks were completed in this thread pool.	Tasks/sec	Compare the value of this measure across the thread pools to figure out the thread pool that has completed the maximum number of tasks per second.
Current_block_tasks	Indicates the number of tasks that are currently blocked in this thread pool.	Number	A high value for this measure indicates that there are no more threads in the thread pool to service the tasks. To avoid the tasks from being blocked, administrators can calibrate the thread pool to accommodate enough threads to service the tasks.
Pending_tasks	Indicates the number of tasks that are pending in this thread pool.	Number	Compare the value across the thread pools to identify the thread pool on which the maximum number of tasks are pending. A sudden/gradual increase in the measure is an indication for the administrators to add additional threads to the thread pool.
Total_block_tasks	Indicates the rate at which tasks were blocked in this thread pool during the last measurement period.	Tasks/sec	A low value is desired for this measure.