kubernetes - Kubernetes ClusterIP not balancing load across Pods
问题描述
I am having an issue with load balancing of traffic between communicating Deployment resources deployed in Kubernetes.
Namely the Deployment A communicates to Deployment B via vanilla ClusterIP type Service. When Deployment A starts causing large load on Deployment B, the Deployment B starts scaling new pods, but when new Pods become ready, they get no traffic.
What seems to be happening is that ClusterIP Service is balancing incoming connections instead of incoming requests, so that only pods which were available before scale-up are under load.
My question is then:
How to make ClusterIP balance the incoming requests across Pods, instead of balancing incoming TCP connections? In case that such thing is not possible, what is the alternative for balancing the HTTP traffic between two deployments?
解决方案
推荐阅读
- bash - 将具有相同名称但扩展名不同的文件保留在curent目录中,并将剩余文件移动到新的director.y
- r - 如何通过向量化以及来自 dplyr 的 group_by 函数迭代列
- dataframe - 如何将数据框值转换为 Map[String,List[String]]?
- regex - 来自“已接收”电子邮件标题的正则表达式匹配时间
- jestjs - 开玩笑找不到测试匹配
- python - ValueError:检查目标时出错:预期dense_2有2维,但得到的数组形状为(1,)
- python - 如何根据条件在熊猫数据框的多列上分配值
- javascript - 如何在节点请求中包含凭据?
- angular - 在 Angular 中正确实现“保存表单并关闭”按钮
- android - 设备旋转后双向 DataBinding 出错