Best Practice Using STS with S3 Client

[shirady]

HI,
I have some parts of my code related to STS and I want to understand the best practice when using STS and S3 Client.

1. When creating a new STS (using AWS SDK V2), when migrating to AWS SDK V3 should we add a region to the configuration? (as it is in S3 client, if so - how do I know which region?)

const AWS = require('aws-sdk'); 
...
const sts = new AWS.STS(); //should we add region here?

2. After using assumeRoleWithWebIdentity we get in the response the Credentials - What is the best practice? As I see it in our code, we create a new S3 client, but I wonder if it would be better to configure the existing S3 Client?

Thanks,
Shira

[shirady]

@RanVaknin would you please share your opinion?
You wrote in the past (copied from here) that

"... each client will maintain its own connections and sockets. When you create multiple clients you can quickly run into connection timeout issues as your Node.js application's connection pool is limited."

But also:

"... we create the s3 client once for specific credentials ...".

In the case of an S3 client, its credentials are generated from the STS, so should we update the config or create a new S3 client every time?

Thanks,
Shira