1. mongos run with: --setParameter connPoolMaxShardedConnsPerHost=10
2. use 24 mongo to connect mongos, 24 connections to mongos
3. run command, such as:

db.example.insert({'foo': 6})

from the mongos log, i found:

2017-12-18T12:01:48.525+0800 I NETWORK [conn21] Ending idle connection to host mongo1260/10.160.188.108:27017,10.160.189.108:27017,10.160.243.108:27017(with timeout of 0 seconds) because the pool meets constraints; 13 connections to that host remain open
2017-12-18T12:01:48.526+0800 I NETWORK [conn11] Ending idle connection to host mongo1260/10.160.188.108:27017,10.160.189.108:27017,10.160.243.108:27017(with timeout of 0 seconds) because the pool meets constraints; 12 connections to that host remain open

but I found there are still 22 connections from mongos to mongod(10.160.243.108), netstat command as below:

netstat -antp | grep ESTABLISHED | fgrep '10.160.243.108:27017' | awk '{print $5}' | sort | uniq -c

the 22 connections will always keep established. the output as below:

22 10.160.243.108:27017

so, may I have a wrong use on connPoolMaxShardedConnsPerHost ?

official docs about connPoolMaxShardedConnsPerHost:

Set the maximum size of the connection pools for communication to the shards. The size of a pool does not prevent the creation of additional connections, but does prevent the connection pools from retaining connections above this limit.