peilun-conflux

#1261 fixes one locking issue.

@yangzhe1990 Another performance bottleneck is the `BTreeMap` in `node_ref_maps`. It consumes more CPU resources in `get_cache_info` now for a state containing 1 million accounts. I recall that before adding `DeltaMptId`...

BTW, changing `BTreeMap` to `hashbrown::HashMap` increases the tps from 5500 to 6000, and still introduces a significant overhead.

This can be configured in the independent pos configuration file. By default it's `pos_config/pos_config.yaml`. The corresponding configuration entry is `base.data_dir`. A commented-out line should have been provided in the file.

The current implementation of broadcasting ServeHeader at phase change is not working. We should probably exchange this in handshaking or do it periodically.

If the warning is not very frequent and your node is making progress, your node should be okay, and this is more likely an issue of other peers. A node...

Are you using the exact version `v2.0.0`? The hardcoded hardfork height in that version has been delayed since `v2.0.0-fix` and `v2.0.0` should not be used. It's more suggested to use...

No mining is possible until you node syncs to the latest height. 在 2022年7月24日，03:10，anthonypan08 ***@***.***> 写道：  Any ideas/updates? — Reply to this email directly, view it on GitHub, or...

Now try to rerun with `additional_maintained_snapshot_count` set to verify it's related. Another guess is that this is a random error because a newly created sqlite3 database may not be ready...

Setting `additional_maintained_snapshot_count` does not necessary cause this issue.