The invention discloses a container cluster delayed shrinkage scheduling method and
system and belongs to the container cluster shrinkage scheduling field. The method comprises the following steps: 1) an agent monitoring module monitors in real time
performance index and health status data of a container bearing a service and sends the result to a scheduling module; 2) the scheduling module judges whether the reported
performance index and the health status data surpass preset threshold values, if so, sends an instruction to the agent monitoring module to create a new container, or otherwise, judges whether there is a newly-created container, and if there is the newly-created container, turn to the step 3); and 3) the scheduling module judges whether the
performance index is smaller than the preset threshold value, if so, a load balancer stops allocating service traffic, and after the data is processed, the scheduling module issues an instruction to delete the new container, or otherwise, turn to the step 1). The whole
system is prevented from the detect of
data loss of an existing mode through
delay and TCP resetting, thereby ensuring that the cases of
packet loss and brief interruption of services are not caused in the load balancing configuration update process to the maximum degree.