C# How I can upload file to MinIO (AWS S3 compatible API) via gRPC without buffering data? - amazon-s3

Question

How can I upload large files to MinIO (AWS S3 compatible API) via gRPC service without buffering data?
I have gRPC service with following definition:
service MediaService {
rpc UploadMedia(stream UploadMediaRequest) returns (UploadMediaResponse);
}
message UploadMediaRequest {
oneof Data {
UploadMediaMetadata metadata = 1;
UploadMediaStream fileStream = 2;
}
}
message UploadMediaMetadata {
string bucket = 1;
string virtialDirectory = 2;
string fileName = 3;
string contentType = 4;
map<string, string> attributes = 6;
}
message UploadMediaStream {
bytes bytes = 1;
}
And implementation of UploadMedia:
public override async Task<UploadMediaResponse> UploadMedia(
IAsyncStreamReader<UploadMediaRequest> requestStream,
ServerCallContext context)
{
UploadMediaMetadata? metadata = null;
var token = context.CancellationToken;
var traceId = context.GetHttpContext().TraceIdentifier;
await using var memoryStream = new MemoryStream();
await foreach (var req in requestStream.ReadAllAsync(token))
{
if (req.DataCase == UploadMediaRequest.DataOneofCase.Metadata)
{
metadata = req.Metadata;
_logger.LogTrace("[Req: {TraceId}] Received metadata", traceId);
}
else
{
await memoryStream.WriteAsync(req.FileStream.Bytes.Memory, token);
_logger.LogTrace("[Req: {TraceId}] Received chunk of bytes", traceId);
}
}
if (metadata == null)
{
throw new RpcException(new Status(StatusCode.InvalidArgument, "Not found metadata."));
}
memoryStream.Seek(0L, SeekOrigin.Begin);
var uploadModel = _mapper.Map<UploadModel>(metadata);
uploadModel.FileStream = memoryStream;
var file = await _fileService.UploadFile(uploadModel, token);
await _eventsService.Notify(new MediaUploadedEvent(file.PublicId), token);
_logger.LogTrace("[Req: {TraceId}] File uploaded", traceId);
return new UploadMediaResponse { File = _mapper.Map<RpcFileModel>(file) };
}
At the method I read request stream and write data to MemoryStream. After that I upload file to storage:
var putObjectArgs = new PutObjectArgs()
.WithStreamData(fileStream)
.WithObjectSize(fileStream.Length)
.WithObject(virtualPath)
.WithBucket(bucket)
.WithContentType(contentType)
.WithHeaders(attributes);
return _storage.PutObjectAsync(putObjectArgs, token);
I want to upload files without buffering data in Memory.
I think I can write bytes from stream to disk and after that create FileStream, but I don't want one more dependency.