rwkarg

The latest reference source for .NET Framework doesn't appear to have the fix: https://github.com/microsoft/referencesource/blob/5697c29004a34d80acdaf5742d7e699022c64ecd/mscorlib/system/threading/Tasks/TPLETWProvider.cs#L520-L522

We saw the same issue with 3.1.7

We run in a multi-cluster scenario (for the ability to destroy/recreate individual clusters without impact) and I don't know that this can be addressed outside of a full scheduling manager....

A hill-climbing algorithm like is used for the [CLR Thread Pool](https://mattwarren.org/2017/04/13/The-CLR-Thread-Pool-Thread-Injection-Algorithm/) could be a candidate for this. It basically will add threads (or instances in this case) and see if...

This would be great for ensuring that only impactful scaling happens with Kafka consumers, especially since _all_ values between p/2 and p-1 are essentially noop for scaling (they all have...

the Kafka scaler already fakes the metrics to cap the instance count at the partition count. https://github.com/kedacore/keda/blob/28d0b69b124ab6a28c7fe736b47aa3246b4f37ee/pkg/scalers/kafka_scaler.go#L452-L457 Could similarly fake the metric value to result in the desired instance count....

It's essentially `totalLag = desiredInstances * s.metadata.lagThreshold` The existing code do a max where `desiredInstances` == the partition count, but if we figure out that the break points are for...

Haven't had any time to look at this, but it would be a nice improvement.

@js8080 No, it does not: https://github.com/open-telemetry/opentelemetry-dotnet/issues/2054 If you're currently using OpenTracing/Jaeger -> Jaeger Collector then you have three options to migrate to OpenTelemetry (all involving standing up new infrastructure): 1....

We similarly have a wrapper around RabbitMQ.Client we've been using for a few years (that exposes consumers/exchanges as `ISource/TargetBlock`) which we instrumented for OpenTracing and are looking at moving to...