block transaction replay right after block update gives errors

[chris-bce]

System information

Erigon version: ./erigon --version

erigon version 2.45.1-stable-83398871

OS & Version: Linux

Chain/Network: erigon mainnet

Expected behaviour

replaying block transactions immediately after block update should succeed

Actual behaviour

replaying block transactions immediately after block update shows errors. adding a time.sleep(1) after getting the update and before replaying block transactions seems to make them go away

Steps to reproduce the behaviour

use the following script:

from typing import Optional

import json
import pprint
import subprocess

current = None
next_id = 1


def pop_id() -> int:
    global next_id
    result = next_id
    next_id += 1
    return result


def poll_for_new_block(prev: Optional[str]) -> str:
    while True:
        s = subprocess.check_output("""curl -X POST -H "Content-Type: application/json" --data '{"jsonrpc":"2.0","method":"eth_getBlockByNumber","params":["latest", false],"id":%d}' localhost:8545""" % pop_id(), shell=True, stderr=subprocess.DEVNULL)
        #print(s)
        data = json.loads(s)
        block_hash = data['result']['hash']
        if prev is None or block_hash != prev:
            return block_hash

current = poll_for_new_block(None)

while True:
    current = poll_for_new_block(current)
    data = json.loads(subprocess.check_output("""curl -X POST -H "Content-Type: application/json" --data '{"jsonrpc":"2.0","method":"trace_replayBlockTransactions","params":["%s", ["stateDiff"]],"id":%d}' localhost:8545""" % (current, pop_id()), shell=True, stderr=subprocess.DEVNULL))
    pprint.pprint(data)

Backtrace

there will be a stream of errors in the output like:

{'error': {'code': -32000,
           'message': 'first run for txIndex 0 error: nonce too low: address '
                      '0x1aCa6E70c0aA9A23a340B0F4308eb0dE520dBd9D, tx: 18 '
                      'state: 19'},
 'id': 772,
 'jsonrpc': '2.0'}
{'error': {'code': -32000,
           'message': 'first run for txIndex 0 error: nonce too low: address '
                      '0xD6f859ba5076e7690b2056f697e5097f9FE836Ed, tx: 9483 '
                      'state: 9484'},
 'id': 881,
 'jsonrpc': '2.0'}
 {'error': {'code': -32000,
           'message': 'first run for txIndex 0 error: nonce too low: address '
                      '0xd78ceBa1f9291dadE3539bD558eB5D9B969ec66F, tx: 13213 '
                      'state: 13214'},
 'id': 1034,
 'jsonrpc': '2.0'}

[bgelb]

This seems to be due to a regression introduced somewhere after release 2.42.

I see the failure with 2.45.1, and I don't see it on 2.42 (which is the prior version deployed on my systems).

[AskAlexSharov]

hmm... it's already 2nd similar issue (first #7686)
i will try to check - maybe we don't call "commit" after staged sync cycle.
in-theory we send notification after commit - so all data must be visible for readers.

[AskAlexSharov]

oookey... seems I understand: Released v2.45.2

[github-actions]

This issue is stale because it has been open for 40 days with no activity. Remove stale label or comment, or this will be closed in 7 days.

[github-actions]

This issue was closed because it has been stalled for 7 days with no activity.