Open patrick-ogrady opened 3 years ago
Previously, I was making 32
concurrent fetches. I've reduced this to 8
to see if it fixes the issue. Even with such a small concurrency, populating the trace response is consuming all CPU resources (on a 48 core box):
When running
rosetta-ethereum
on Ropsten, I ran across an error where I could not get traces for certain blocks:This is 99% likely to be caused by too short of a trace timeout and/or
geth
response timeout (where response time is elevated due to concurrent fetches).