How to set limit for pool connection

[pragya92]

Please answer these questions before submitting your issue. Thanks!

What version of Cassandra are you using?

4.x

What version of Gocql are you using?

v1.6.0

What version of Go are you using?

1.19

What did you do?

I am creating a pool connection with cluster.NumConns = 20. so that it can create only 20 connection/pod.
container := dbType + "." + dbName
username := String(container+".username", "")
password := String(container+".password", "")
host := String(container+".host", "")
db := String(container+".db", "")
dc := String(container+".dc", "")
port1 := String(container+".port", "")

// fmt.Println(container, " ", username, " ", password, " ", host, " ", db, " ", dc, " ", port1)
port, err := strconv.Atoi(port1)

HostList := strings.Split(host, ",")

cluster := gocql.NewCluster(HostList...)
pass := gocql.PasswordAuthenticator{username, password, nil}
cluster.Keyspace = db
cluster.Authenticator = pass
cluster.Consistency = gocql.LocalQuorum
//cluster.SerialConsistency = gocql.LocalSerial
if err == nil {
	cluster.Port = port // Default port
}

if dbName == "cass_comp_gcp" || dbName == "cass_comp_gcp_newIN" {
	cluster.DisableInitialHostLookup = true
	cluster.SocketKeepalive = 10 * time.Second
}
cluster.NumConns = 20
//
cluster.PoolConfig.HostSelectionPolicy = gocql.DCAwareRoundRobinPolicy(dc)
cluster.HostFilter = gocql.DataCentreHostFilter(dc)
cluster.Timeout = 5 * time.Second
if GetMode() != "prod" {
	cluster.ConnectTimeout = time.Second * 20
	//cluster.DisableInitialHostLookup = true
}
// cluster.SocketKeepalive = 10 * time.Second

return cluster.CreateSession()

What did you expect to see?

We are using the above function in our read API which is capable of serving around 5,000 requests per second. Our goal is to limit the number of connections created per pod to 20, even if there is an increase in request traffic. Although we have added a limit of 20 connections, we have observed that if the request count increases, it opens more than 20 connections, and some of them remain open even after the normal pace.

What did you see instead?

It is creating around 200-600 connections per pod, which is not getting closed even after the normal pace. This is causing a load issue for Cassandra.

If you are having connectivity related issues please share the following additional information

Describe your Cassandra cluster

please provide the following information

• output of nodetool status
• output of SELECT peer, rpc_address FROM system.peers
• rebuild your application with the gocql_debug tag and post the output