立即与支持人员聊天
与支持团队交流

Foglight 7.1.0 - Performance Tuning Field Guide

Overview Hardware and Operating System Tuning Management Server Tuning Java Virtual Machine Tuning Backend Database Tuning High Availability (HA) Tuning Agent Tuning Store and Forward Processing Appendix: Performance Health Check Appendix: Analyzing a Support Bundle

Appendix: Performance Health Check

The purpose of a health check is to quickly ensure that Foglight® is functioning properly. There are many things you can check. The more important ones are documented in this chapter.

•
Is the Server Getting the Right Amount of Memory?
•
Is the Server Getting Enough CPU?
•
Is the Database too Slow?
•
Is the Database Growth Reasonable?
•
Is There Too Much Data to Process?
•
Is the Model Stable?
•
Are Too Many Alarms Firing?
•
Is the Business Logic Properly Tuned?
•
Are There Too Many User Requests?

Is the Server Getting the Right Amount of Memory?

By default, Foglight® tries to reserve three quarters of the local available physical memory. It is normal for Foglight to consume all of that memory on start up. Foglight reserves its full amount of memory on start up. Foglight is configured this way to avoid excess garbage collection.

Foglight uses a lot of temporary memory. This is due to the metric processing. It is normal and healthy for Foglight’s internal memory consumption to rise and fall.

If there is enough memory reserved, the Foglight JavaTM Virtual Machine (JVM) heap graph will display an adequate amount of free memory. It is normal for the memory to cycle up and down. As long as these memory cycles are not close together (which indicates excessive garbage collection), then all is well. Also, as long as free memory continues to be returned through garbage collection, then the amount of memory configured is fine.

Management Server Memory is Healthy if ...

•
The amount of memory consumed is typically less than 80 percent of the maximum.
•
There are no log messages indicating a failure to create threads.
•
The operating system is not constantly paging to keep the Management Server’s memory available.

Management Server Memory Checks

•
Check the -Xmx and -Xms settings in the Foglight® log to make sure they match expectations.
•
Look at the JVM heap graph and make sure the free memory is not stuck beyond 80 percent of the maximum memory.
•
Look for a failure to create native threads in the log. If native threads cannot be created, then either stack space or total memory should be decreased.
•
Look at the Management Server system’s paging. If memory is frequently paging, then the memory requested by the server is too much for the system. This could be because another large process is running.
相关文档

The document was helpful.

选择评级

I easily found the information I needed.

选择评级