What is the difference between connectTimeoutMS, socketTimeoutMS and maxTimeMS ?

Setting Default Value client.connect Description
connectTimeoutMS 30000 The connectTimeoutMS sets the number of milliseconds a socket stays inactive before closing during the connection phase of the driver. That is to say, when the application initiates a connection, when a replica set connects to new members, or when a replica set reconnects to members. A value of 10000 milliseconds would mean the driver would wait up to 10 seconds for a response from a MongoDB server.
socketTimeoutMS 360000 The socketTimeoutMS sets the number of milliseconds a socket stays inactive after the driver has successfully connected before closing. If the value is set to 360000 milliseconds, the socket closes if there is no activity during a 6 minutes window.
maxTimeMS N/A The maxTimeMS setting specifies how long MongoDB should run an operation before cancelling it. If the maxTimeMS is set to 10000 milliseconds, any operation that runs over that limit returns a timeout error.

Fail fast during connection

In this scenario, the developer wants to ensure that the driver does not hang during the connection phase or spend an unnecessarily long time attempting to connect to replica set members who are not reachable.

As a general rule you should ensure that the connectTimeoutMS setting is not lower than the longest network latency you have to a member of the set. If one of the secondary members is on the other side of the planet and has a latency of 10000 milliseconds, setting the connectTimeoutMS to anything lower will prevent the driver from ever connecting to that member.

socketTimeoutMS as a way to abort operations

Developers sometimes try to use socketTimeoutMS to end operations which may run for too long and slow down the application, but doing so may not achieve the intended result.

Closing the socket forces a reconnect of the driver’s connection pool and introduces latency to any other operations which are queued up. Chronically slow operations will therefore cause a large number of reconnect requests, negatively impacting throughput and performance.

Also, closing the socket does not terminate the operation; it will continue to run on the MongoDB server, which could cause data inconsistencies if the application retries the operation on failure.

That said, there are some important use cases for socketTimeoutMS. It’s possible that a MongoDB process may error out, or that a misconfigured firewall may close a socket connection without sending a FIN packet. In these cases there is no way for the driver to detect that the connection has died, and socketTimeoutMS is essential to ensure that the sockets are closed correctly.

A good rule of thumb is to set socketTimeoutMS to two to three times the length of the slowest operation which runs through the driver.

socketTimeoutMS and large connection pools

Having a large connection pool does not always reduce reconnection requests. Consider the following example: an application has a connection pool size of 5 sockets and has socketTimeoutMS set to 5000 milliseconds. Operations occur, on average, every 3000 milliseconds, and reconnection requests are frequent. Each socket times out after 5000 milliseconds, which means that all sockets must do something during that 5000 millisecond period to avoid closing. One message every 3000 milliseconds is not enough to keep the sockets active, so several of the sockets will time out after 5000 milliseconds.

Reducing the pool size to 1 will fix the problem.

The special meaning of 0

Setting connectTimeoutMS and socketTimeoutMS to the value 0 has a special meaning. It causes the application to use the operating system’s default socket timeout value.

maxTimeMS is the option you are looking for

Many developers set a low socketTimeoutMS value, intending to prevent long-running server operations from slowing down the application. maxTimeMS is usually a better choice; it allows MongoDB itself to cancel operations which run for more than maxTimeMS milliseconds.

The following example demonstrates how to use MaxTimeMS with a find operation.

// Execute a find command
col.find({"$where": "sleep(100) || true"})
  .maxTimeMS(50)
  .count(function(err, count) {
});

What does the keepAlive setting do?

keepAlive is a socket setting available from Node.js that in theory will keep a socket alive by sending periodic probes to MongoDB. However, this only works if the operating system supports SO_KEEPALIVE, and still might not work if a firewalls ignores or drops the keepAlive packets.

On misconfigured firewalls

Internal firewalls which exist between application servers and MongoDB are often misconfigured, and are overly aggressive in their culling of socket connections. If you experience unexpected network behavior, here are some things to check:

  1. The firewall should send a FIN packet when closing a socket, allowing the driver to detect that the socket is closed.
  2. The firewall should allow keepAlive probes.

I’m getting ECONNRESET when calling client.connect

This can occur if the connection pool is too large.

const client = new MongoClient('mongodb://localhost:27017/test?maxPoolSize=5000');
client.connect(function(err) {
	// connection
});

If this operation causes an ECONNRESET error, you may have run into the file descriptor limit for your Node.js process.

In most operating systems, each socket connection is associated with a file descriptor. Many operating systems have a limit on how many such file descriptors can be used by a single process.

The way to fix the descriptor limit issue is to increase the number of file descriptors for the Node.js process. On Mac OS and Linux you do this with the ulimit shell command.

ulimit -n 6000

This sets the maximum number of file descriptors for the process to 6000, allowing Node.js to connect with a pool size of 5000 sockets.

How can I prevent a slow operation from delaying other operations?

Note

The driver is only affected by the slow train operations if the number of slow operations is larger than the max pool size.

While Node.js is asynchronous, MongoDB is not. Currently, MongoDB uses a single execution thread per socket. This means that it will only execute a single operation on a socket at any given point in time. Any other operations sent to that socket will have to wait until the current operation is finished. If you have a slow-running operation which holds up other operations, the best solution is to create a separate connection pool for the slow operation, isolating it from other, faster operations.

Ensure your connection string is valid for Replica Set

The connection string passed to the driver MUST use the fully qualified host names for the servers as set in the replicaset config. Given the following configuration settings for your replicaset.

{
	"_id" : "testSet",
	"version" : 1,
	"protocolVersion" : 1,
	"members" : [
		{
			"_id" : 1,
			"host" : "server1:31000",
		},
		{
			"_id" : 2,
			"host" : "server2:31001",
		},
		{
			"_id" : 3,
			"host" : "server3:31002",
		}
	]
}

You must ensure server1, server2 and server3 are resolvable from the driver for the Replicaset discovery and failover to work correctly.