Help to understand "SQL Connections" in CRDB dashboard

Hi, I just discovered this metric in CRDB dashboard. From the beginning of the crdb stack setup, the metric increased gradually from 20 to 32 in 3 days. Although during the 3 days, sometimes there is no interaction with the crdb stack.
Don’t quite understand why this metric never dropped. If it needs to manually disconnect, why so many existing connections did not cause failure when huge incoming requests came in. Is there a mechanism that crdb automatically detect which connection is actually in idle and disconnect it and assign to new requests?

Hey @derek.z

Let’s get to the bottom of this so that can optimize the number of connections. First, does the application that is connecting to the clusters have a built in timeout on its end as well? This can determine if the SQL connection to the DB is being controlled from that side. Secondly, are these other connections active still, ie running some kind of transaction or query? And finally, please be aware that depending on the view in the AdminUI, you may be looking at the SQL connections for the entire cluster. If there is a load balancer in front of the cluster, it may make multiple connections for each transactionto help balance workload.

Let me know if this additional information can be provided.

Cheers,
Ricardo