Secondary read on replica set


having a replica set with 3 nodes (2 secondary nodes) does it make sense to use secondary read preference? I was thinking that using secondary read preference in that case will increase the read throughput as the load should be spread between these two secondary nodes. Is that the case?

Best regards

Yes but at the high risk of reading stale data. For analytic purposes, it is fine. But for end user that also performs updates, it is a big NO.

So if it’s OK for the end user to get stale data then this can be considered a good practice, right?

I would not consider providing stale data a good practice. An acceptable one in some use-cases.

Hi @Lukasz_Brygala,

You can check this documentation for Read Preference Use cases:

Please feel free to reach out if you have any questions.

Kind Regards,

Why . due to stale data or some other reason ?

Indeed, it is entirely due to stale data.