Docs Menu

Docs HomeDevelop ApplicationsMongoDB Manual

Tuning Your Connection Pool Settings

On this page

  • Calculate Maximum Number of Connections
Problem
Solution
Slow application-side operation times that are not reflected in the database server logs or the real time panel.

Use connectTimeoutMS to ensure the driver does not wait indefinitely during the connection phase.

Set connectTimeoutMS to a value greater than the longest network latency you have to a member of the set.

For example: if a member has a latency of 10000 milliseconds, setting connectTimeoutMS to 5000 (milliseconds) prevents the driver from connecting to that member.

A misconfigured firewall closes a socket connection incorrectly and the driver cannot detect that the connection closed improperly.

Use socketTimeoutMS to ensure that sockets are always closed.

Set socketTimeoutMS to two or three times the length of the slowest operation that the driver runs.

The server logs or real time panel show that the application spends too much time creating new connections.

Not enough connections are available at startup. Allocate connections in the pool by setting minPoolSize.

Set minPoolSize to the number of connections you want to be available at startup.

The MongoClient instance ensures that number of connections exists at all times.

The load on the database is low and there's a small number of active connections at any time. Application performs fewer operations at any one time than expected.
Increase maxPoolSize, or increase the number of active threads in your application or the framework you are using.
Database CPU usage is higher than expected. The server logs or real time panel show more connection attempts than expected.
Decrease the maxPoolSize or reduce the number of threads in your application. This can reduce load and response times.

Warning

Do not use socketTimeoutMS to prevent long-running server operations. Instead, use maxTimeMS() with queries so that the server can cancel long-running operations.

Calculate usage to find the number of operations running for each connection.

Consider an application server connecting to a replica set with three members. In this scenario, the application server creates a connection pool for each replica set member. You also need to consider the connections each client, on each node, creates for monitoring.

The maximum number of connections that are opened by the application server is based on the maxPoolSize, the number of members, and each member's monitoring and RTT connections.

For example, you can calculate the outgoing connections from an application to a three-member replica set with the following formula:

(
**100** {:urioption:`maxPoolSize`}
+ **1** monitoring connection
+ **1** RTT connection
)
* 3 replica set members
= **306** outgoing connections from the application.
←  Connection Pool OverviewDisable Transparent Huge Pages (THP) →