-
Notifications
You must be signed in to change notification settings - Fork 4.9k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[API Proposal]: [HTTP/3] Support multiple HTTP/3 connections #101535
Comments
Tagging subscribers to this area: @dotnet/ncl |
A customer has run into this problem - grpc/grpc-dotnet#2404 They have more than 100 concurrent requests and this limitation is preventing new requests from being sent. |
Is I imagine whoever wants multiple connections for HTTP/2 would also want multiple for HTTP/3. A property that sets both would be good. |
We cannot rename the existing
I will mention you suggestion during API review as an alternative to get the feel for it. |
Looks good as proposed namespace System.Net.Http;
public sealed partial class SocketsHttpHandler : System.Net.Http.HttpMessageHandler
{
public bool EnableMultipleHttp3Connections { get { throw null; } set { } }
} |
Should it be added to WinHttpHandler has |
At the moment, we do not support H/3 on |
Background and motivation
Similarly to HTTP/2, we want to introduce support for multiple HTTP/3 connections.
Both HTTP/2 and HTTP/3 have stream limits. As RFC Bidirectional Streams suggests, the minimum is 100, which is the current Kestrel default. As #35088 states, this becomes a bottleneck for service-to-service communication, e.g. gRPC.
Implementation depends on #101534.
Original issue: #51775
For comparison with HTTP/2 see the original issue #35088
API Proposal
namespace System.Net.Http; public sealed partial class SocketsHttpHandler : System.Net.Http.HttpMessageHandler { public bool EnableMultipleHttp2Connections { get { throw null; } set { } } + public bool EnableMultipleHttp3Connections { get { throw null; } set { } } }
API Usage
Alternative Designs
N/A
Risks
None. It's an opt-in, default will be
false
, keeping the original behavior in place.The text was updated successfully, but these errors were encountered: