Raising this issue just so we don't lose sight of the ongoing work on this one.
After the block-by-block changes for the datastream writing we noticed that a batch won't seal until we have all responses back from an executor. this saw a significant drop in TPS so we need to continue work on handling this better and not waiting for executors before sealing a batch.
Raising this issue just so we don't lose sight of the ongoing work on this one.
After the block-by-block changes for the datastream writing we noticed that a batch won't seal until we have all responses back from an executor. this saw a significant drop in TPS so we need to continue work on handling this better and not waiting for executors before sealing a batch.