This problem occurred after upgrading the linux kernel in vmware (I used putty to connect to centos7 in the virtual machine in win10). At the beginning, I tried to solve it by the second method of https://blog.csdn.net/googchar/article/details/8190104 : In the
above figure, both check boxes are kept selected, and Seconds between keepalives I set to 600.
The result is still not working. Then I tried to modify the configuration file /etc/ssh/sshd_config of the ssh server (in my case, centos7), and made the following modifications:
TCPKeepAlive yes
ClientAliveInterval 600
ClientAliveCountMax 3
Seems ok