I encountered this behaviour many times in many servers which processed lots of network connections.
# ss -nt state time-wait sport ne :80 and sport ne :10050 | sort -k3
0 0 127.0.0.1:13530 127.0.0.1:8888
0 0 127.0.0.1:21978 127.0.0.1:8080
0 0 127.0.0.1:32490 127.0.0.1:8080
0 0 127.0.0.1:42922 127.0.0.1:8080
0 0 127.0.0.1:50728 127.0.0.1:8080
0 0 127.0.0.1:51542 127.0.0.1:8888
0 0 127.0.0.1:6274 127.0.0.1:8888
0 0 127.0.0.1:65264 127.0.0.1:8888
0 0 172.16.40.100:10000 172.16.40.5:3010
0 0 172.16.40.100:10002 172.16.40.34:3010
0 0 172.16.40.100:10002 172.16.40.97:3020
0 0 172.16.40.100:10004 172.16.40.116:3010
0 0 172.16.40.100:10004 172.16.40.21:3010
0 0 172.16.40.100:10008 172.16.40.30:3010
0 0 172.16.40.100:10010 172.16.40.216:3020
0 0 172.16.40.100:10012 172.16.40.30:3010
0 0 172.16.40.100:10014 172.16.40.131:3010
0 0 172.16.40.100:10014 172.16.40.22:3010
0 0 172.16.40.100:10014 172.16.40.33:3010
This is a part of ss output. As you may see, there are several strings with duplicated outgoing time_wait sockets. Such as:
0 0 172.16.40.100:10002 172.16.40.34:3010
0 0 172.16.40.100:10002 172.16.40.97:3020
or
0 0 172.16.40.100:10014 172.16.40.131:3010
0 0 172.16.40.100:10014 172.16.40.22:3010
0 0 172.16.40.100:10014 172.16.40.33:3010
I googled this question but could not get a reasonable explanation of this topic.
Thanks a lot!
The lines in this display are connections, not sockets. There are exactly zero 'duplicated sockets' here. There is a duplicated port, because at the server end the port is always the same. However either the client IP address or the client port is always different. Or both.