A method and device for preventing pod memory overflow in a container orchestration framework

Abstract

The present invention provides a method and device for preventing POD memory overflow in a container orchestration framework. The method includes the following steps: monitoring the current memory size of the POD, and calculating the memory usage rate according to the current memory size and a preset memory threshold. When the memory usage rate is greater than the set threshold, a request for flow control is sent; after receiving the request for flow control, the interception level will be judged according to the number of consecutive requests, and the packet loss operation will be performed according to the packet loss rate corresponding to the interception level to adjust the flow; When the traffic has been adjusted with the highest interception level within the set time period, a memory threshold adjustment request is sent; after receiving the memory threshold adjustment request, the application memory threshold is dynamically changed to reduce the memory usage. It avoids the possibility that the application will be killed due to business surge, abnormal access or application GC causing excessive memory usage, resulting in memory overflow.

Description

technical field [0001] The invention relates to the technical field of POD applications in a kubernetes cluster, in particular to a method and a device for preventing POD memory overflow in a container orchestration framework. Background technique [0002] POD applications run in Kubernetes clusters. Kubernetes is a container orchestration framework open sourced by Google, also called a container orchestration engine. It supports automated deployment, large-scale scalability, and application containerized management. When deploying an application in a production environment, it is usually necessary to deploy multiple instances of the application in order to load balance application requests. When the memory used by the POD (POD, application executor and operating environment) during the running process exceeds the set value, It will be killed by the controller, causing the application to be unable to provide services in a short time. Since the memory resources used by the ap...

AI Technical Summary

Problems solved by technology

When deploying an application in a production environment, it is usually necessary to deploy multiple instances of the application to load balance the application requests. When the memory used during the operation of the POD (POD, application execution program and operating environment) exceeds the set value, It will be killed by the controller, causing the application to be unable to provide services in a short period of time. Since the memory resources periodically used by the application cannot be quickly tested, there may be two situations in which the preset limit is set. The first is that the actual business demand is always greater than the limit value , another situation is that there is still a large distance between the actual business and the limit value

In both cases, there will be losses

Images