I'm currently running load tests on my local parachain which includes Frontier integration for EVM compatibility. During these tests, I'm sending thousands of transactions via the RPC endpoint. Observations show that these transactions remain in the pending state in the txpool. However, as they begin to get processed, only about hundreds transactions are included in each block.
Upon inspecting the blocks, I notice that only 50% of the block weight is being utilized.
It's worth noting that I haven't done any benchmarking yet, which I thought might increase the likelihood of fully utilizing the block capacity. Actually I want to check that correct benchmarking would avoid the hypothetical bad situation caused because of load testing.
Could I be missing something in my approach?