Image explanation of TKE-Optimized series

Last updated: 2020-04-10 16:04:28

PDF

Overview

TKE-Optimized images are images of Kubernetes cluster nodes produced by Tencent Kubernetes Engine (TKE) based on TencentOS-kernel. Fully optimized for TKE, these images have higher security and stability. In particular, the images provide a custom kernel that is maintained by the Tencent Cloud team and can be updated by using hotpatches. We recommend that you select TKE-Optimized series images to deploy your TKE cluster nodes.

Features

FeatureDescription
Kernel Customization
  • Based on kernel 4.14.105, which has been long supported by the kernel community.
  • Adds new features applicable to Tencent Cloud scenarios to optimize the performance of the kernel and fix major defects.
Container Support
  • Enhances isolation and performance for TKE.
  • Supports isolating system files such as meminfo, vmstat, cpuinfo, stat, and loadavg.
  • Supports isolating sysctl interfaces, such as tcp_no_delay_ack and tcp_max_orphans.
  • Fixed many files system and network bugs.
Performance Optimization
  • Optimizes the xfs memory allocation to resolve the xfs kmem_alloc failure.
  • Optimizes memory allocation while receiving network packets, to prevent excessive memory usage when a large number of UDP packets are received.
  • Limits the memory usage of system page caches, to prevent the lack of memory from affecting service performance and causing the out-of-memory (OOM) error.
Support and Updates
  • Regularly updates the kernel to enhance security and functionality.
  • Updates the kernel by using hotpatches.