I have a MongoDB server (4.4.19) that has a M40 Tier and maximum of 3000 connections.
In the past few weeks, the connections have been filled up and we had no other choice but to Test Resilience and failover to a secondary node to get free connections again.
But as soon as we have available connections, they are filled up pretty quick and in less than a day, it goes up to 2700 and the cycle continues.
We want to find out what are these connections filling up the server and work with the appropriate clients and limit their connection parameters.
My question is: How do we find out these consuming connections that are filling up the available connection slots? We would like to have the output by clients names (application names) rather than IPs. It would be great to have the top 10 consuming connections (e.g.: If 1 particular client, has 500+ connections, it would be #1 )
Any input is greatly appreciated.
Thank you!
Hi @Juan_Polanco - Welcome to the community.
Have you checked the logs for the cluster to help possibly narrow down where the issue connection(s) are coming from?
The Atlas in-app chat support may be able to assist you with this in terms of providing the application / driver’s connecting if you can specify approximately the time frames when it occurs although you may need to double check with them. Alternatively, you could raise a support case if you have a support subscription (above Basic Support level) as well.
The following page may be of use to you regarding connection issues.
Regards,
Jason