Running the RPC adapter
Instructions on how to locally run the RPC adapter
Running the RPC adapter allows for interaction with the Acala EVM+, be it with the local development network, public test network or either of the main networks.
There are three ways of running the RPC node in a local environment:
Running the RPC adapter as a package allows you to use the RPC adapter without having to set up anything else. To run it in a this way, use:
npx @acala-network/eth-rpc-adapter \
--endpoint ws://localhost:9944 \
--local-mode
Flags can be appended to the command in order to adapt the default configuration to your needs. Flags are documented at the bottom of this page.
git clone --recurse-submodules [email protected]:AcalaNetwork/bodhi.js.git
Once the repository is cloned install the necessary dependencies as well as build the adapter:
In order to be able to locally build the RPC adapter, you are required to have @microsoft/rush installed.
rush update
rush build -t @acala-network/eth-rpc-adapter
The RPC adapter is ready run:
cd eth-rpc-adapter
yarn start --local-mode [--other-options]
Running the RPC adapter in Docker required Docker to be running and issuing the following command:
docker run -it --rm -e LOCAL_MODE=1 -p 8545:8545 acala/eth-rpc-adapter:v2.5.3 yarn start
Options can be passed to the RPC adapter in three ways:
- 1.As CLI options (recommended)
- 2.As environment variables
- 3.As environment variables in the
.envfile
Passing the options as CLI options is done by appending the values to the command when running the RPC adapter:
npx @acala-network/eth-rpc-adapter -l -e ws://localhost:9944
If you wish to pass the options as environment variables in the terminal, you can add them with the
export statements, or by prepending the command used to spin up RPC adapter:export ENDPOINT_URL=ws://localhost:9944
LOCAL_MODE=1 yarn start
Both the environment variable that is defined by itself as well as the one defined inline with the
yarn start will be used in this case.In case you are building the RPC adapter from source, you can set the options as variables within the
.env file, like thisENDPOIN_URL=ws://localhost:9944
LOCAL_MODE=1
Please don't mix using ENVs and CLI options. CLI options are preferred, and will overwrite ENVs.
More details can also be found by
yarn start --help or npx @acala-network/eth-rpc-adapter --help.ENV | CLI options equivalent | default | explanation |
|---|---|---|---|
ENDPOINT_URL | -e, --endpoint | ws://localhost:9944 | Node websocket endpoint(s): can provide one or more endpoints, seperated by comma url |
SUBQL_URL | --subql | undefined | Subquery url: optional if testing contracts locally that doesn't query logs or historical Tx, otherwise required. more info |
PORT | -p, --port | 8545 | port to listen for http and ws requests |
MAX_CACHE_SIZE | --max-cache-size | 200 | |
MAX_BATCH_SIZE | --max-batch-size | 50 | max batch size for RPC request |
STORAGE_CACHE_SIZE | --max-storage-size | 5000 | max storage cache size |
SAFE_MODE | -s, --safe-mode | 0 | if enabled, TX and logs can only be found after they are finalized |
LOCAL_MODE | -l, --local-mode | 0 | enable this mode when testing with locally running instant-sealing mandala |
RICH_MODE | -r, --rich-mode | 0 | if enabled, default gas params is big enough for most contract deployment and calls, so contract tests from traditional evm world can run unchanged. Note this mode is helpful for testing contracts, but is different than production envionment. more info |
HTTP_ONLY | --http-only | 0 | only allow http requests, disable ws connections |
VERBOSE | -v, --verbose | 1 | print some extra info |
Using the
-e flag and passing the URL of a Mandala, Acala and Karura node allows for running the local RPC adapter while communicating with a public network.In this mode, Txs and logs can only be found after they are finalized. Now deprecated in favor for the
finalized and safe block tags.For local testing, we usually turn this mode on, together with a local
--instant-sealing mandala node. It has some optimization to run faster with local node, and some minor bug prevention.We usually need to specify gas params for some transactions, this is sometimes time-consuming for contract testing, especially with
hardhat where we can't override gas params in config.In rich mode, default gas params are much bigger than normal, so we don't need to worry about overriding gas params when running tests. As a result, all truffle/hardhat tests can be run unchanged.
Rich mode is convenient for contract testing, but we still recommend reading through the gas params and understand how gas works in EVM+, since in prod we might still need to override gas params.
Last modified 3mo ago