副本故障转移后，TCP连接断开连接 [英] TCP Connection gets disconnected after the replica failover

问题描述

我正在使用Availablity Group设置AwlaysOn功能。我在WFC中设置了3个节点，并为两个具有一个主副本和两个辅助副本的数据库设置了三个独立的SQL服务器实例和setup Availablity组。

I'm working on setting up the AwlaysOn feature using Availablity Group. I have setup 3 nodes in WFC and setup three standalone SQL server instances and setup Availablity Group for two databases with one Primary Replica and two secondary replicas.

每当我从辅助副本故障转移到主副本时我与Virtual AG Listener（VIP）的连接在SSMS上断开连接，并且我收到以下错误。

Whenever I failover from Secondary replica to Primary replica my connection to Virtual AG Listener (VIP) is getting disconnected on SSMS and I get the the following error.

将请求发送到服务器时发生传输级错误。

A transport-level error has occurred when sending the request to the server.

现有连接被远程主机强行关闭

这在常规的2008 Cluster实例中是正常的，但我认为它应该在Denali中保持连接，因此它被称为AlwaysON。是否需要调整设置？

This is normal is in a regular 2008 Cluster instance, but I thought it should stay connected in Denali and hence it is called AlwaysON. Is there any settings needs to be adjusted?

推荐答案

Hi Dev！

Hi Dev!

TCP连接在故障转移期间必须断开连接，这就是TCP的工作方式。因此，无论您使用SQL Cluster，Database Mirorring还是AlwaysOn Availability Groups，都需要具有重新连接逻辑。 AlwaysOn的主要观点是你可以用
重新连接相同的名称/ IP，这是你不能轻易用数据库镜像做的事情，而且仍然不需要昂贵的共享磁盘子系统，你不能很容易做集群。所以AlwaysOn本身并没有摆脱所有问题，但是b $ b它是一个非常好的聚类和镜像组合，可以同时为您提供这两种技术的优势。

The TCP connection has to get disconnected during failover, that's the way TCP works. So you need to have a reconnect logic, no matter if you use SQL Cluster, Database Mirorring or AlwaysOn Availability Groups. The major point in AlwaysOn is that you can just reconnect with the same name/IP, a thing you couldn't easily do with Database Mirroring, and still don't need an expensive shared disk subsystem, which you couldn't easily do with clustering. So AlwaysOn is not getting rid of all problems per se, but it is a quite nice combination of clustering and mirroring, giving you the advantages of both technologies at once.

干杯

路西法

这篇关于副本故障转移后，TCP连接断开连接的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！