Go to file
2022-08-23 18:48:27 +00:00
.cargo keep RUSTFLAGS in one place 2022-05-17 16:36:42 +00:00
.vscode minimal app served its purpose 2022-05-20 02:01:02 +00:00
bin less verbose 2022-07-25 19:04:11 +00:00
config load the redirected page from config 2022-08-12 19:08:28 +00:00
entities upgrade 2022-08-21 08:30:43 +00:00
fifomap cache db data in a map 2022-08-10 02:37:43 +00:00
linkedhashmap drop redis-cell in favor of simpler query 2022-08-15 22:50:56 +00:00
migration upgrade 2022-08-21 08:30:43 +00:00
redis-rate-limit multiple ways to sign 2022-08-19 20:18:12 +00:00
web3_proxy dry redis connections and use bearer tokens 2022-08-23 18:48:27 +00:00
wrk move data files 2022-05-06 01:40:43 +00:00
.dockerignore dash consistency 2022-08-06 05:46:33 +00:00
.env move dev to different ports 2022-08-06 05:49:52 +00:00
.gitignore setup database and stub migrations 2022-07-26 04:55:07 +00:00
Cargo.lock dry redis connections and use bearer tokens 2022-08-23 18:48:27 +00:00
Cargo.toml drop redis-cell in favor of simpler query 2022-08-15 22:50:56 +00:00
docker-compose.common.yml setup volatile redis 2022-08-16 05:00:29 +00:00
docker-compose.prod.yml setup volatile redis 2022-08-16 05:00:29 +00:00
docker-compose.yml setup volatile redis 2022-08-16 05:00:29 +00:00
Dockerfile dash consistency 2022-08-06 05:46:33 +00:00
LICENSE add LICENSE 2022-03-04 19:56:05 -08:00
README.md polish docs 2022-08-16 21:10:03 +00:00
TODO.md move no unwrap todo to v1 2022-08-19 21:09:03 +00:00

web3_proxy

Web3_proxy is a fast caching and load balancing proxy for web3 (Ethereum or similar) JsonRPC servers.

Under construction! This code is under active development. The basics seem to work, but theres lots of tests and features to write still.

Signed transactions (eth_sendRawTransaction) are sent in parallel to the configured private RPCs (eden, ethermine, flashbots, etc.).

All other requests are sent to an RPC server on the latest block (alchemy, moralis, rivet, your own node, or one of many other providers). If multiple servers are in sync, they are prioritized by active_requests/soft_limit. Note that this means that the fastest server is most likely to serve requests and slow servers are unlikely to ever get any requests.

Each server has different limits to configure. The soft_limit is the number of parallel active requests where a server starts to slow down. The hard_limit is where a server starts giving rate limits or other errors.

$ cargo install sea-orm-cli
$ cargo run --release -- --help
   Compiling web3_proxy v0.1.0 (/home/bryan/src/web3_proxy/web3_proxy)
    Finished release [optimized + debuginfo] target(s) in 17.69s
     Running `target/release/web3_proxy --help`
Usage: web3_proxy [--port <port>] [--workers <workers>] [--config <config>]

web3_proxy is a fast caching and load balancing proxy for web3 (Ethereum or similar) JsonRPC servers.

Options:
  --port            what port the proxy should listen on
  --workers         number of worker threads
  --config          path to a toml of rpc servers
  --help            display usage information

Start the server with the defaults (listen on http://localhost:8544 and use ./config/example.toml which proxies to a bunch of public nodes:

cargo run --release -- --config ./config/example.toml

Common commands

Check that the proxy is working:

curl -X POST -H "Content-Type: application/json" --data '{"jsonrpc":"2.0","method":"web3_clientVersion","id":1}' 127.0.0.1:8544

Check that the websocket is working:

$ websocat ws://127.0.0.1:8544

{"id": 1, "method": "eth_subscribe", "params": ["newHeads"]}

{"id": 2, "method": "eth_subscribe", "params": ["newPendingTransactions"]}

{"id": 3, "method": "eth_subscribe", "params": ["newPendingFullTransactions"]}

{"id": 4, "method": "eth_subscribe", "params": ["newPendingRawTransactions"]}

You can copy config/example.toml to config/production-$CHAINNAME.toml and then run docker-compose up --build -d start proxies for many chains.

Database entities

This command only needs to be run during development. Production should use the already generated entities.

When developing new database migrations, after you migrate, run this command to generate updated entity files. It's best to keep the migration and entity changes in the same commit.

sea-orm-cli generate entity -u mysql://root:dev_web3_proxy@127.0.0.1:13306/dev_web3_proxy -o entities/src

After running the above, you will need to manually fix some columns: Vec<u8> -> sea_orm::prelude::Uuid and i8 -> bool. Related: https://github.com/SeaQL/sea-query/issues/375 https://github.com/SeaQL/sea-orm/issues/924

Flame Graphs

Flame graphs make a developer's join of finding slow code painless:

$ cat /proc/sys/kernel/kptr_restrict
1
$ echo 0 | sudo tee /proc/sys/kernel/kptr_restrict
0
$ CARGO_PROFILE_RELEASE_DEBUG=true cargo flamegraph

GDB

Developers can run the proxy under gdb for advanced debugging:

cargo build --release && RUST_LOG=web3_proxy=debug rust-gdb --args target/debug/web3_proxy --listen-port 7503 --rpc-config-path ./config/production-eth.toml

TODO: also enable debug symbols in the release build by modifying the root Cargo.toml

Load Testing

Test the proxy:

wrk -s ./data/wrk/getBlockNumber.lua -t12 -c400 -d30s --latency http://127.0.0.1:8544
wrk -s ./data/wrk/getLatestBlockByNumber.lua -t12 -c400 -d30s --latency http://127.0.0.1:8544

Test geth (assuming it is on 8545):

wrk -s ./data/wrk/getBlockNumber.lua -t12 -c400 -d30s --latency http://127.0.0.1:8545
wrk -s ./data/wrk/getLatestBlockByNumber.lua -t12 -c400 -d30s --latency http://127.0.0.1:8545

Test erigon (assuming it is on 8945):

wrk -s ./data/wrk/getBlockNumber.lua -t12 -c400 -d30s --latency http://127.0.0.1:8945
wrk -s ./data/wrk/getLatestBlockByNumber.lua -t12 -c400 -d30s --latency http://127.0.0.1:8945

Note: Testing with getLatestBlockByNumber.lua is not great because the latest block changes and so one run is likely to be very different than another.

Run ethspam and versus for a more realistic load test:

ethspam --rpc http://127.0.0.1:8544/u/someuserkey | versus --concurrency=100 --stop-after=10000 http://127.0.0.1:8544/u/someuserkey