mirror of
https://gitlab.com/pulsechaincom/erigon-pulse.git
synced 2024-12-23 12:07:17 +00:00
8a877c227a
* Expose C Handle of lmdbTx * LoadExecutionFunctionPointer * silkworm ExecuteBlocks * linter * CLI flag for SilkwormExecutionFunc * linter * Call SilkwormExecutionFunc in SpawnExecuteBlocksStage * Commit transaction after Silkworm execution * Fix batch initialization * Make batch_size uint64_t instead of size_t * max_block in silkworm_execute_blocks * More accurate logProgress * Use CHandle exposed by lmdb-go * Enable Silkworm on Linux only * Update lmdb-go * Error when attempting to use Silkworm not on Linux * Move SilkwormFlag from cmd/utils/flags.go to turbo/cli/flags.go * Integrate Silkworm into cmd/integration * Check against ReaderBuilder/WriterBuilder/ChangeSetHook in Silkworm execution * Refactor SpawnExecuteBlocksStage * linters * Small clean-ups * Move ChangeSetHook inside executeBlockWithGo * No need to tamper with batch size in executeBlocksWithSilkworm * fix * Fix * Fixup Co-authored-by: Alexey Sharp <alexeysharp@Alexeys-iMac.local> |
||
|---|---|---|
| .. | ||
| commands | ||
| main.go | ||
| Readme.md | ||
Integration - tool to run TurboGeth stages in custom way: run single stage, or run all stages but reorg every X blocks, etc...
By examples.
All commands require parameter --chaindata=/path/to/chaindata - I will skip it for readability.
integration --help
integration print_stages --block=1_000_000
# run all stages (which do not need internet).
integration state_stages --unwind=1 --unwind_every=10 # process 10 blocks, then reorg 1 block, then process 10 blocks, ...
integration state_stages --unwind=10 --unwind_every=1 # process 1 block, then reorg 10 blocks, then process 1 blocks, ...
# Run single stage
integration stage_senders
integration stage_exec
integration stage_exec --block=1_000_000 # stop at 1M block
integration stage_hash_state
integration stage_ih
integration stage_history
integration stage_tx_lookup
# Drop data of single stage
integration stage_exec --reset
integration stage_history --reset
...
# reset all data after stage_senders
integration reset_state
# hack which allows to force clear unwind stack of all stages
clear_unwind_stack
The way I usually run it:
go run -trimpath ./cmd/integration state_stages --chaindata=/path/to/chaindata --unwind=10 --unwind_every=20 --pprof
Pre-requirements of state_stages command:
- Headers/Bodies must be downloaded
- TxSenders stage must be executed