## Changes
### Add `pallet-grandpa-benchmarking` crate
Upstream `pallet-grandpa` does not benchmark `report_equivocation`, only
the raw `check_equivocation_proof` crypto proxy.
This PR adds a new `pallet-grandpa-benchmarking` wrapper crate (modelled
after `pallet-session-benchmarking`) that benchmarks both
`report_equivocation` and `note_stalled` against the DataHaven runtimes,
using upstream `check_equivocation_proof()`.
### Add node's benchmark pallet subcommand to benchmarks script
Running the bench for `report_equivocation` requires a real ed25519
verifier. We typically use `frame-omni-bencher`, but this helper
executes the runtime as a WASM blob. In that environment the ed25519
host function does not work as a real verifier, and since the
equivocation proof contains real signatures created outside the WASM
sandbox, the verification always fails and the extrinsic returns
`InvalidEquivocationProof`. So, benchmarks for `pallet_grandpa` must run
via the node's `benchmark pallet` subcommand.
The `run-benchmarks.sh` script was updated to support this case running
directly via node. Any palet included in the new `NODE_PALLETS` list,
routes selected pallets through the node binary, while all other pallets
continue to use `frame-omni-bencher`.
### Calculate proper weights for production
Weight files for all three runtimes (stagenet, testnet, mainnet) are
updated with the
new `report_equivocation(v, n)` linear weight function derived from real
measurements.
---------
Co-authored-by: Steve Degosserie <723552+stiiifff@users.noreply.github.com>
Co-authored-by: Ahmad Kaouk <56095276+ahmadkaouk@users.noreply.github.com>
* Add `run_benchmarks.sh` script to run runtime benchmarks
* Sets up benchmark configs and directory structure to store weights
(`operator/runtime/<RUNTIME>/weights`)
* (naive) fixes to some benchmarks:
* `pallet_datahaven_native_transfer`:
* use a mock for `NativeTokenId`
* look at the balance difference of the treasury instead of the total
(this makes the benchmark agnostic to genesis setup)
* `snowbridge_pallet_system` / `snowbridge_pallet_system_v2` use native
token xcm location vs relay chain one. Add missing benchmark methods and
update fixture with valid data.
* `snowbridge_pallet_ethereum_client`: update fixtures with valid data
* `snowbrige_pallet_inbound_queue_v2`: set EthereumGatewayAddress when
initializing storage on benchmark and use a mock message processor ( as
fixture has `CreateAsset` payload which is not supported in the
`EigenLayerMessageProcessor`)
* `snowbridge_pallet_outbound_queue_v2`: add missing
`submit_delivery_receipt` benchmark which required a dedicated fixture
(all copied from the upstream pallet)
* `pallet_treasury`: Use an `ExistentialDeposit` of `1` on benchmark,
else payout fails.
* `pallet_transaction_payment`: Use a custom `WeightToFee` that makes
the Fee small, else account in benchmark cannot pay for fees (It is
funded a multiplier of `ExistentialDeposit` and is expected for that to
be enough, but it's not in our particular setup).
* comment out `pallet_identity` and `pallet_im_online` due to
incompatibilities (to be addressed later)
* Basic benchmark run to set `WeightInfo` from `weights` in configs
(real run should be done later using target hardware)
---------
Co-authored-by: Ahmad Kaouk <ahmadkaouk.93@gmail.com>
Co-authored-by: Tobi Demeco <50408393+TDemeco@users.noreply.github.com>
Co-authored-by: undercover-cactus <lola@moonsonglabs.com>
Co-authored-by: TDemeco <tdemeco@itba.edu.ar>
