Liangying.Wei

Why "disconnection" event for one SignalR client will be handled by **all** instances?

Are you using `Default` mode or `Classic` mode? For `Classic` mode, it is possible that some clients connected to serverless mode.  For now, there is no API exposed to...

Please email me `lianwei@microsoft` your instance name and the time period when the abnormal spikes happened for me to dump the connections.

> Could it be that the Azure SignalR Service is overloaded? Sounds possible What Unit is your resource? FYI here is a performance case study for tiny group https://docs.microsoft.com/en-us/azure/azure-signalr/signalr-concept-performance#small-group You...

A similar approach for webpubsub is here : https://github.com/Azure/azure-rest-api-specs/pull/15907/commits/40a3903f9529d1c852f2b92d03802f90fd5ae060

Looks causing by https://github.com/dotnet/aspnetcore/issues/32307 that in 6.0 [ConnectionContext](https://github.com/dotnet/aspnetcore/blob/main/src/Servers/Connections.Abstractions/src/BaseConnectionContext.cs#L26) this `IFeatureCollection Features` is considered to be in assembly `Microsoft.Extension.Features`  However `WebSocketConnectionContext` implements `Features` in assembly `Microsoft.AspNetCore.Http.Features`  As a result,...

There is an on-going effort to fix this issue as tracked in https://github.com/dotnet/aspnetcore/issues/38699

Sorry for the late response. When using AzureSignalR, blazor client requests are going through Azure SignalR to the server instead of triggering the server's HTTP middleware anymore. So the way...

Thanks for pointing it out. We need to update the javascript sample to have a custom retry policy, e.g. have a random delay before every try.

Also add more details on how to add/change trace listeners in tsg