Distributed task scheduling platform XXL-JOB
XXL-JOB is a lightweight distributed task scheduling framework. Its core design goals are rapid development, easy learning, light weight, and easy expansion. Now open source and connected to the online product lines of many companies, out of the box.
I did some tests when investigating xxl-job (hereinafter referred to as xxl), and now I present the results of the initial test.
When using xxl, first look at his architecture diagram, as follows
The first time I saw it, I couldn't figure out the flow of the process, so I just tried to explore it directly.
The determinant of whether xxl can be used in a production environment must be stability, and stability is the overriding factor. In this test, we will start with stability.
My test base environment is window10 professional, i7+16g, and the maximum memory of tomcat8.0.35 is increased to 1g.
Ok, start the configuration of xxl, first configure 4 actuators, as shown below
Each executor corresponds to a tomcat instance.
Create 4 new tasks in task management
Each task adopts the GLUE method, and only outputs "XXL-JOB, Hello World." on the actuator side. The 
latter three are also performed according to this configuration. After all the configurations are completed, the test execution starts for 24 hours.
The result is as shown in the figure above. The program ran for about 24 hours, with about 345,000 scheduling times, without a single failure. But I found a problem, first look at the picture, the following picture shows the memory usage of Tomcat
The memory usage of the four executors tomcat shown in the figure above is not high. Within 100m, the memory usage of the initial startup is about 250m, but the memory usage of the tomcat of xxl-admin is abnormal, which has reached 1288m, and the initial memory usage is about 750m, and It is still growing. I also consulted the author @雪里 on this issue. The author said that he is not clear about the problem. I have a bold guess. It may be caused by the heartbeat link or the communication link being too occupied and unable to be released in time?
This test did not test the failover of xxl-admin, but only tested the stability, for the next cruel test, 5 calls per second, each executor failover on two machines, to see how long xxl can last, let us wait and see.