When running on the weak staging machines I noticed that a highish volume (~2k rps) of a cheap call such as eth_blockNumber was causing a big increase in cpu usage. We should figure out if it the increased usage is focused on a single point and if it is, try opmizing it.
Also the baseline cpu usage of followers is much higher than the miner, so it'd be worthwile taking a look why that is too.
When running on the weak staging machines I noticed that a highish volume (~2k rps) of a cheap call such as
eth_blockNumberwas causing a big increase in cpu usage. We should figure out if it the increased usage is focused on a single point and if it is, try opmizing it.Also the baseline cpu usage of followers is much higher than the miner, so it'd be worthwile taking a look why that is too.