Check Runtime
Describes the try-runtime command-line tool for testing a specified runtime state against a production snapshot of chain state.
Last updated
Describes the try-runtime command-line tool for testing a specified runtime state against a production snapshot of chain state.
Last updated
The try-runtime command-line tool enables you to query a snapshot of runtime storage using an in-memory-externalities data structure to store state. By using the in-memory storage, you can write tests for a specified runtime state so that you can test against real chain state before going to production.
In its simplest form, you can use try-runtime to test the runtime state by doing the following:
Connect to a remote node.
Call a specific runtime API.
Retrieve state from the node at a specific block.
Write tests for the data retrieved.
The initial motivation for try-runtime came from the need to test runtime changes against state from a real chain. Prior TestExternalities and BasicExternalities existed for writing unit and integrated tests with mock data, but lacked the ability to test against a chain's actual state. The try-runtime tool extends TestExternalities and BasicExternalities by retrieving state using the following RPC endpoints for the node:
rpc_get_storage
rpc_get_keys_paged
(see remote externalities lib for more details.)
After using the key-value database to retrieve state, try-runtime inserts the data into TestExternalities.
The try-runtime tool has its own implementation of externalities called remote_externalities which is just a wrapper around TestExternalities that uses a generic key-value store where data is type encoded.
The diagram below illustrates the way externalities sits outside a compiled runtime as a means to capture the storage of that runtime.
With remote_externalities, you can capture some chain state and run tests on it. Essentially, RemoteExternalities will populate a TestExternalities with a real chain's data.
To query state, try-runtime uses the RPC methods provided by StateApiClient. In particular:
storage This method returns the storage value for the key that represents the block you specify.
storage_key_paged This method returns the keys that match a prefix you specify with pagination support.
The most common use case for try-runtime is to help you prepare for storage migration and runtime upgrades.
Because the RPC calls that query storage are computationally expensive, there are a number of command-line options you should set for a running node before you use the try-runtime command. To prepare a node for try-runtime testing, set the following options:
Set --rpc-max-payload 1000 to ensure large RPC queries can work.
Set --rpc-cors all to ensure WebSocket connections can come through.
You can combine try-runtime with fork-off-substrate to test your chain before production. Use try-runtime to test your chain's migration and its pre and post states. Then, use fork-off-substrate if you want to check that block production continues after the migration.
By default, runtime upgrade hooks—which can be defined inside of the runtime or inside pallets—specify what should happen when there's been a runtime upgrade. That is, the default on_runtime_upgrade method only describes runtime state after the upgrade. However, it is possible to use methods provided by try-runtime to inspect and compare the runtime state before and after a runtime upgrade for testing purposes.
If you enable the try-runtime feature for the runtime, you can define pre-upgrade and post-upgrade hooks for the runtime as follows:
With these function, you can use the pre_upgrade hook to retrieve the runtime state and return it as a Vec result. You can the pass the Vec as input parameter to the post_upgrade hook.
To use try-runtime from the command-line, run your node with the --features=try-runtime flag.
For example:
You can use the following subcommands with try-runtime:
on-runtime-upgrade: Executes tryRuntime_on_runtime_upgrade against the given runtime state.
offchain-worker: Executes offchainWorkerApi_offchain_worker against the given runtime state.
execute-block: Executes core_execute_block using the given block and the runtime state of the parent block.
follow-chain: Follows a given chain's finalized blocks and applies to all its extrinsics. This allows the behavior of a new runtime to be inspected over a long period of time, with real transactions coming as input.
To view usage information for a specific try-runtime subcommand, specify the subcommand and the --help flag. For example, to see usage information for try-runtime on-runtime-upgrade, you can run the following command:
For example, you can run try-runtime with the on-runtime-upgrade subcommand for a chain running locally with a command like this:
You can use try-runtime to re-execute code from the ElectionProviderMultiPhase offchain worker on localhost:9944 with a command like this:
You can run the migrations of the local runtime on the state of SomeChain with a command like this:
You can run try-runtime against the state for a specific block number with a command like this:
Notice that this command requires the --no-spec-name-check command-line option.
