🔧 Chore
#596 has implemented benchmark testing and profiling for the tally process. Now we hope to implement something similar for the RPC server to understand why it at times slows down significantly.
It may be difficult to implement this programmatically due to the complicated setup process involving multiple components. Our goal here is to get profiling data that shows where the server is spending time once a request is accepted. CometBFT's rpc/core package is a good candidate for where to start profiling / tracing.
A broader goal would be to understand if there is a bottleneck in the lower level HTTP server handling and whether RPC request congestion can result from other chain processes like heavy block execution.
🔧 Chore
#596 has implemented benchmark testing and profiling for the tally process. Now we hope to implement something similar for the RPC server to understand why it at times slows down significantly.
It may be difficult to implement this programmatically due to the complicated setup process involving multiple components. Our goal here is to get profiling data that shows where the server is spending time once a request is accepted. CometBFT's
rpc/corepackage is a good candidate for where to start profiling / tracing.A broader goal would be to understand if there is a bottleneck in the lower level HTTP server handling and whether RPC request congestion can result from other chain processes like heavy block execution.