Customer is facing queue full exception and IO timeout issue

Message from Customer :

We have recently been experiencing an issue with our .NET web api where the Couchbase connection gets into a queue full state and is only recoverable with a service restart. We receive a fair amount of req's/second, on the order of 100's, each of which does an Upsert to Couchbase with a small payload. There is no unusual server or network load. It seems that once the issue occurs it cascades into an unrecoverable state and does not automatically recover. We are using the latest .NET libraries (3.3.2) with the DI Extension library (3.3.0) on .NET 6.0 and Windows server 2022. We have also tried scaling our cluster which did not seem to make a difference. Maybe it's the way we are acquiring the bucket references? We are currently using this method in each request:
var cache = await _cacheBucketProvider.GetBucketAsync().ConfigureAwait(false);