You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
916 lines
40 KiB
916 lines
40 KiB
#!/usr/bin/env python3
|
|
# Copyright (c) 2014-2021 The Bitcoin Core developers
|
|
# Distributed under the MIT software license, see the accompanying
|
|
# file COPYING or http://www.opensource.org/licenses/mit-license.php.
|
|
"""Base class for RPC testing."""
|
|
|
|
import configparser
|
|
from enum import Enum
|
|
import argparse
|
|
import logging
|
|
import os
|
|
import pdb
|
|
import random
|
|
import re
|
|
import shutil
|
|
import subprocess
|
|
import sys
|
|
import tempfile
|
|
import time
|
|
|
|
from typing import List
|
|
from .address import create_deterministic_address_bcrt1_p2tr_op_true
|
|
from .authproxy import JSONRPCException
|
|
from . import coverage
|
|
from .p2p import NetworkThread
|
|
from .test_node import TestNode
|
|
from .util import (
|
|
MAX_NODES,
|
|
PortSeed,
|
|
assert_equal,
|
|
check_json_precision,
|
|
get_datadir_path,
|
|
initialize_datadir,
|
|
p2p_port,
|
|
wait_until_helper,
|
|
)
|
|
|
|
|
|
class TestStatus(Enum):
|
|
PASSED = 1
|
|
FAILED = 2
|
|
SKIPPED = 3
|
|
|
|
TEST_EXIT_PASSED = 0
|
|
TEST_EXIT_FAILED = 1
|
|
TEST_EXIT_SKIPPED = 77
|
|
|
|
TMPDIR_PREFIX = "bitcoin_func_test_"
|
|
|
|
|
|
class SkipTest(Exception):
|
|
"""This exception is raised to skip a test"""
|
|
|
|
def __init__(self, message):
|
|
self.message = message
|
|
|
|
|
|
class BitcoinTestMetaClass(type):
|
|
"""Metaclass for BitcoinTestFramework.
|
|
|
|
Ensures that any attempt to register a subclass of `BitcoinTestFramework`
|
|
adheres to a standard whereby the subclass overrides `set_test_params` and
|
|
`run_test` but DOES NOT override either `__init__` or `main`. If any of
|
|
those standards are violated, a ``TypeError`` is raised."""
|
|
|
|
def __new__(cls, clsname, bases, dct):
|
|
if not clsname == 'BitcoinTestFramework':
|
|
if not ('run_test' in dct and 'set_test_params' in dct):
|
|
raise TypeError("BitcoinTestFramework subclasses must override "
|
|
"'run_test' and 'set_test_params'")
|
|
if '__init__' in dct or 'main' in dct:
|
|
raise TypeError("BitcoinTestFramework subclasses may not override "
|
|
"'__init__' or 'main'")
|
|
|
|
return super().__new__(cls, clsname, bases, dct)
|
|
|
|
|
|
class BitcoinTestFramework(metaclass=BitcoinTestMetaClass):
|
|
"""Base class for a bitcoin test script.
|
|
|
|
Individual bitcoin test scripts should subclass this class and override the set_test_params() and run_test() methods.
|
|
|
|
Individual tests can also override the following methods to customize the test setup:
|
|
|
|
- add_options()
|
|
- setup_chain()
|
|
- setup_network()
|
|
- setup_nodes()
|
|
|
|
The __init__() and main() methods should not be overridden.
|
|
|
|
This class also contains various public and private helper methods."""
|
|
|
|
def __init__(self):
|
|
"""Sets test framework defaults. Do not override this method. Instead, override the set_test_params() method"""
|
|
self.chain: str = 'regtest'
|
|
self.setup_clean_chain: bool = False
|
|
self.nodes: List[TestNode] = []
|
|
self.network_thread = None
|
|
self.rpc_timeout = 60 # Wait for up to 60 seconds for the RPC server to respond
|
|
self.supports_cli = True
|
|
self.bind_to_localhost_only = True
|
|
self.parse_args()
|
|
self.disable_syscall_sandbox = self.options.nosandbox or self.options.valgrind
|
|
self.default_wallet_name = "default_wallet" if self.options.descriptors else ""
|
|
self.wallet_data_filename = "wallet.dat"
|
|
# Optional list of wallet names that can be set in set_test_params to
|
|
# create and import keys to. If unset, default is len(nodes) *
|
|
# [default_wallet_name]. If wallet names are None, wallet creation is
|
|
# skipped. If list is truncated, wallet creation is skipped and keys
|
|
# are not imported.
|
|
self.wallet_names = None
|
|
# By default the wallet is not required. Set to true by skip_if_no_wallet().
|
|
# When False, we ignore wallet_names regardless of what it is.
|
|
self.requires_wallet = False
|
|
# Disable ThreadOpenConnections by default, so that adding entries to
|
|
# addrman will not result in automatic connections to them.
|
|
self.disable_autoconnect = True
|
|
self.set_test_params()
|
|
assert self.wallet_names is None or len(self.wallet_names) <= self.num_nodes
|
|
if self.options.timeout_factor == 0 :
|
|
self.options.timeout_factor = 99999
|
|
self.rpc_timeout = int(self.rpc_timeout * self.options.timeout_factor) # optionally, increase timeout by a factor
|
|
|
|
def main(self):
|
|
"""Main function. This should not be overridden by the subclass test scripts."""
|
|
|
|
assert hasattr(self, "num_nodes"), "Test must set self.num_nodes in set_test_params()"
|
|
|
|
try:
|
|
self.setup()
|
|
self.run_test()
|
|
except JSONRPCException:
|
|
self.log.exception("JSONRPC error")
|
|
self.success = TestStatus.FAILED
|
|
except SkipTest as e:
|
|
self.log.warning("Test Skipped: %s" % e.message)
|
|
self.success = TestStatus.SKIPPED
|
|
except AssertionError:
|
|
self.log.exception("Assertion failed")
|
|
self.success = TestStatus.FAILED
|
|
except KeyError:
|
|
self.log.exception("Key error")
|
|
self.success = TestStatus.FAILED
|
|
except subprocess.CalledProcessError as e:
|
|
self.log.exception("Called Process failed with '{}'".format(e.output))
|
|
self.success = TestStatus.FAILED
|
|
except Exception:
|
|
self.log.exception("Unexpected exception caught during testing")
|
|
self.success = TestStatus.FAILED
|
|
except KeyboardInterrupt:
|
|
self.log.warning("Exiting after keyboard interrupt")
|
|
self.success = TestStatus.FAILED
|
|
finally:
|
|
exit_code = self.shutdown()
|
|
sys.exit(exit_code)
|
|
|
|
def parse_args(self):
|
|
previous_releases_path = os.getenv("PREVIOUS_RELEASES_DIR") or os.getcwd() + "/releases"
|
|
parser = argparse.ArgumentParser(usage="%(prog)s [options]")
|
|
parser.add_argument("--nocleanup", dest="nocleanup", default=False, action="store_true",
|
|
help="Leave bitcoinds and test.* datadir on exit or error")
|
|
parser.add_argument("--nosandbox", dest="nosandbox", default=False, action="store_true",
|
|
help="Don't use the syscall sandbox")
|
|
parser.add_argument("--noshutdown", dest="noshutdown", default=False, action="store_true",
|
|
help="Don't stop bitcoinds after the test execution")
|
|
parser.add_argument("--cachedir", dest="cachedir", default=os.path.abspath(os.path.dirname(os.path.realpath(__file__)) + "/../../cache"),
|
|
help="Directory for caching pregenerated datadirs (default: %(default)s)")
|
|
parser.add_argument("--tmpdir", dest="tmpdir", help="Root directory for datadirs")
|
|
parser.add_argument("-l", "--loglevel", dest="loglevel", default="INFO",
|
|
help="log events at this level and higher to the console. Can be set to DEBUG, INFO, WARNING, ERROR or CRITICAL. Passing --loglevel DEBUG will output all logs to console. Note that logs at all levels are always written to the test_framework.log file in the temporary test directory.")
|
|
parser.add_argument("--tracerpc", dest="trace_rpc", default=False, action="store_true",
|
|
help="Print out all RPC calls as they are made")
|
|
parser.add_argument("--portseed", dest="port_seed", default=os.getpid(), type=int,
|
|
help="The seed to use for assigning port numbers (default: current process id)")
|
|
parser.add_argument("--previous-releases", dest="prev_releases", action="store_true",
|
|
default=os.path.isdir(previous_releases_path) and bool(os.listdir(previous_releases_path)),
|
|
help="Force test of previous releases (default: %(default)s)")
|
|
parser.add_argument("--coveragedir", dest="coveragedir",
|
|
help="Write tested RPC commands into this directory")
|
|
parser.add_argument("--configfile", dest="configfile",
|
|
default=os.path.abspath(os.path.dirname(os.path.realpath(__file__)) + "/../../config.ini"),
|
|
help="Location of the test framework config file (default: %(default)s)")
|
|
parser.add_argument("--pdbonfailure", dest="pdbonfailure", default=False, action="store_true",
|
|
help="Attach a python debugger if test fails")
|
|
parser.add_argument("--usecli", dest="usecli", default=False, action="store_true",
|
|
help="use bitcoin-cli instead of RPC for all commands")
|
|
parser.add_argument("--perf", dest="perf", default=False, action="store_true",
|
|
help="profile running nodes with perf for the duration of the test")
|
|
parser.add_argument("--valgrind", dest="valgrind", default=False, action="store_true",
|
|
help="run nodes under the valgrind memory error detector: expect at least a ~10x slowdown. valgrind 3.14 or later required. Forces --nosandbox.")
|
|
parser.add_argument("--randomseed", type=int,
|
|
help="set a random seed for deterministically reproducing a previous test run")
|
|
parser.add_argument('--timeout-factor', dest="timeout_factor", type=float, default=1.0, help='adjust test timeouts by a factor. Setting it to 0 disables all timeouts')
|
|
|
|
group = parser.add_mutually_exclusive_group()
|
|
group.add_argument("--descriptors", action='store_const', const=True,
|
|
help="Run test using a descriptor wallet", dest='descriptors')
|
|
group.add_argument("--legacy-wallet", action='store_const', const=False,
|
|
help="Run test using legacy wallets", dest='descriptors')
|
|
|
|
self.add_options(parser)
|
|
# Running TestShell in a Jupyter notebook causes an additional -f argument
|
|
# To keep TestShell from failing with an "unrecognized argument" error, we add a dummy "-f" argument
|
|
# source: https://stackoverflow.com/questions/48796169/how-to-fix-ipykernel-launcher-py-error-unrecognized-arguments-in-jupyter/56349168#56349168
|
|
parser.add_argument("-f", "--fff", help="a dummy argument to fool ipython", default="1")
|
|
self.options = parser.parse_args()
|
|
self.options.previous_releases_path = previous_releases_path
|
|
|
|
config = configparser.ConfigParser()
|
|
config.read_file(open(self.options.configfile))
|
|
self.config = config
|
|
|
|
if self.options.descriptors is None:
|
|
# Prefer BDB unless it isn't available
|
|
if self.is_bdb_compiled():
|
|
self.options.descriptors = False
|
|
elif self.is_sqlite_compiled():
|
|
self.options.descriptors = True
|
|
else:
|
|
# If neither are compiled, tests requiring a wallet will be skipped and the value of self.options.descriptors won't matter
|
|
# It still needs to exist and be None in order for tests to work however.
|
|
self.options.descriptors = None
|
|
|
|
def setup(self):
|
|
"""Call this method to start up the test framework object with options set."""
|
|
|
|
PortSeed.n = self.options.port_seed
|
|
|
|
check_json_precision()
|
|
|
|
self.options.cachedir = os.path.abspath(self.options.cachedir)
|
|
|
|
config = self.config
|
|
|
|
fname_bitcoind = os.path.join(
|
|
config["environment"]["BUILDDIR"],
|
|
"src",
|
|
"bitcoind" + config["environment"]["EXEEXT"],
|
|
)
|
|
fname_bitcoincli = os.path.join(
|
|
config["environment"]["BUILDDIR"],
|
|
"src",
|
|
"bitcoin-cli" + config["environment"]["EXEEXT"],
|
|
)
|
|
self.options.bitcoind = os.getenv("BITCOIND", default=fname_bitcoind)
|
|
self.options.bitcoincli = os.getenv("BITCOINCLI", default=fname_bitcoincli)
|
|
|
|
os.environ['PATH'] = os.pathsep.join([
|
|
os.path.join(config['environment']['BUILDDIR'], 'src'),
|
|
os.path.join(config['environment']['BUILDDIR'], 'src', 'qt'), os.environ['PATH']
|
|
])
|
|
|
|
# Set up temp directory and start logging
|
|
if self.options.tmpdir:
|
|
self.options.tmpdir = os.path.abspath(self.options.tmpdir)
|
|
os.makedirs(self.options.tmpdir, exist_ok=False)
|
|
else:
|
|
self.options.tmpdir = tempfile.mkdtemp(prefix=TMPDIR_PREFIX)
|
|
self._start_logging()
|
|
|
|
# Seed the PRNG. Note that test runs are reproducible if and only if
|
|
# a single thread accesses the PRNG. For more information, see
|
|
# https://docs.python.org/3/library/random.html#notes-on-reproducibility.
|
|
# The network thread shouldn't access random. If we need to change the
|
|
# network thread to access randomness, it should instantiate its own
|
|
# random.Random object.
|
|
seed = self.options.randomseed
|
|
|
|
if seed is None:
|
|
seed = random.randrange(sys.maxsize)
|
|
else:
|
|
self.log.debug("User supplied random seed {}".format(seed))
|
|
|
|
random.seed(seed)
|
|
self.log.debug("PRNG seed is: {}".format(seed))
|
|
|
|
self.log.debug('Setting up network thread')
|
|
self.network_thread = NetworkThread()
|
|
self.network_thread.start()
|
|
|
|
if self.options.usecli:
|
|
if not self.supports_cli:
|
|
raise SkipTest("--usecli specified but test does not support using CLI")
|
|
self.skip_if_no_cli()
|
|
self.skip_test_if_missing_module()
|
|
self.setup_chain()
|
|
self.setup_network()
|
|
|
|
self.success = TestStatus.PASSED
|
|
|
|
def shutdown(self):
|
|
"""Call this method to shut down the test framework object."""
|
|
|
|
if self.success == TestStatus.FAILED and self.options.pdbonfailure:
|
|
print("Testcase failed. Attaching python debugger. Enter ? for help")
|
|
pdb.set_trace()
|
|
|
|
self.log.debug('Closing down network thread')
|
|
self.network_thread.close()
|
|
if not self.options.noshutdown:
|
|
self.log.info("Stopping nodes")
|
|
if self.nodes:
|
|
self.stop_nodes()
|
|
else:
|
|
for node in self.nodes:
|
|
node.cleanup_on_exit = False
|
|
self.log.info("Note: bitcoinds were not stopped and may still be running")
|
|
|
|
should_clean_up = (
|
|
not self.options.nocleanup and
|
|
not self.options.noshutdown and
|
|
self.success != TestStatus.FAILED and
|
|
not self.options.perf
|
|
)
|
|
if should_clean_up:
|
|
self.log.info("Cleaning up {} on exit".format(self.options.tmpdir))
|
|
cleanup_tree_on_exit = True
|
|
elif self.options.perf:
|
|
self.log.warning("Not cleaning up dir {} due to perf data".format(self.options.tmpdir))
|
|
cleanup_tree_on_exit = False
|
|
else:
|
|
self.log.warning("Not cleaning up dir {}".format(self.options.tmpdir))
|
|
cleanup_tree_on_exit = False
|
|
|
|
if self.success == TestStatus.PASSED:
|
|
self.log.info("Tests successful")
|
|
exit_code = TEST_EXIT_PASSED
|
|
elif self.success == TestStatus.SKIPPED:
|
|
self.log.info("Test skipped")
|
|
exit_code = TEST_EXIT_SKIPPED
|
|
else:
|
|
self.log.error("Test failed. Test logging available at %s/test_framework.log", self.options.tmpdir)
|
|
self.log.error("")
|
|
self.log.error("Hint: Call {} '{}' to consolidate all logs".format(os.path.normpath(os.path.dirname(os.path.realpath(__file__)) + "/../combine_logs.py"), self.options.tmpdir))
|
|
self.log.error("")
|
|
self.log.error("If this failure happened unexpectedly or intermittently, please file a bug and provide a link or upload of the combined log.")
|
|
self.log.error(self.config['environment']['PACKAGE_BUGREPORT'])
|
|
self.log.error("")
|
|
exit_code = TEST_EXIT_FAILED
|
|
# Logging.shutdown will not remove stream- and filehandlers, so we must
|
|
# do it explicitly. Handlers are removed so the next test run can apply
|
|
# different log handler settings.
|
|
# See: https://docs.python.org/3/library/logging.html#logging.shutdown
|
|
for h in list(self.log.handlers):
|
|
h.flush()
|
|
h.close()
|
|
self.log.removeHandler(h)
|
|
rpc_logger = logging.getLogger("BitcoinRPC")
|
|
for h in list(rpc_logger.handlers):
|
|
h.flush()
|
|
rpc_logger.removeHandler(h)
|
|
if cleanup_tree_on_exit:
|
|
shutil.rmtree(self.options.tmpdir)
|
|
|
|
self.nodes.clear()
|
|
return exit_code
|
|
|
|
# Methods to override in subclass test scripts.
|
|
def set_test_params(self):
|
|
"""Tests must override this method to change default values for number of nodes, topology, etc"""
|
|
raise NotImplementedError
|
|
|
|
def add_options(self, parser):
|
|
"""Override this method to add command-line options to the test"""
|
|
pass
|
|
|
|
def skip_test_if_missing_module(self):
|
|
"""Override this method to skip a test if a module is not compiled"""
|
|
pass
|
|
|
|
def setup_chain(self):
|
|
"""Override this method to customize blockchain setup"""
|
|
self.log.info("Initializing test directory " + self.options.tmpdir)
|
|
if self.setup_clean_chain:
|
|
self._initialize_chain_clean()
|
|
else:
|
|
self._initialize_chain()
|
|
|
|
def setup_network(self):
|
|
"""Override this method to customize test network topology"""
|
|
self.setup_nodes()
|
|
|
|
# Connect the nodes as a "chain". This allows us
|
|
# to split the network between nodes 1 and 2 to get
|
|
# two halves that can work on competing chains.
|
|
#
|
|
# Topology looks like this:
|
|
# node0 <-- node1 <-- node2 <-- node3
|
|
#
|
|
# If all nodes are in IBD (clean chain from genesis), node0 is assumed to be the source of blocks (miner). To
|
|
# ensure block propagation, all nodes will establish outgoing connections toward node0.
|
|
# See fPreferredDownload in net_processing.
|
|
#
|
|
# If further outbound connections are needed, they can be added at the beginning of the test with e.g.
|
|
# self.connect_nodes(1, 2)
|
|
for i in range(self.num_nodes - 1):
|
|
self.connect_nodes(i + 1, i)
|
|
self.sync_all()
|
|
|
|
def setup_nodes(self):
|
|
"""Override this method to customize test node setup"""
|
|
extra_args = [[]] * self.num_nodes
|
|
if hasattr(self, "extra_args"):
|
|
extra_args = self.extra_args
|
|
self.add_nodes(self.num_nodes, extra_args)
|
|
self.start_nodes()
|
|
if self.requires_wallet:
|
|
self.import_deterministic_coinbase_privkeys()
|
|
if not self.setup_clean_chain:
|
|
for n in self.nodes:
|
|
assert_equal(n.getblockchaininfo()["blocks"], 199)
|
|
# To ensure that all nodes are out of IBD, the most recent block
|
|
# must have a timestamp not too old (see IsInitialBlockDownload()).
|
|
self.log.debug('Generate a block with current time')
|
|
block_hash = self.generate(self.nodes[0], 1, sync_fun=self.no_op)[0]
|
|
block = self.nodes[0].getblock(blockhash=block_hash, verbosity=0)
|
|
for n in self.nodes:
|
|
n.submitblock(block)
|
|
chain_info = n.getblockchaininfo()
|
|
assert_equal(chain_info["blocks"], 200)
|
|
assert_equal(chain_info["initialblockdownload"], False)
|
|
|
|
def import_deterministic_coinbase_privkeys(self):
|
|
for i in range(self.num_nodes):
|
|
self.init_wallet(node=i)
|
|
|
|
def init_wallet(self, *, node):
|
|
wallet_name = self.default_wallet_name if self.wallet_names is None else self.wallet_names[node] if node < len(self.wallet_names) else False
|
|
if wallet_name is not False:
|
|
n = self.nodes[node]
|
|
if wallet_name is not None:
|
|
n.createwallet(wallet_name=wallet_name, descriptors=self.options.descriptors, load_on_startup=True)
|
|
n.importprivkey(privkey=n.get_deterministic_priv_key().key, label='coinbase', rescan=True)
|
|
|
|
def run_test(self):
|
|
"""Tests must override this method to define test logic"""
|
|
raise NotImplementedError
|
|
|
|
# Public helper methods. These can be accessed by the subclass test scripts.
|
|
|
|
def add_nodes(self, num_nodes: int, extra_args=None, *, rpchost=None, binary=None, binary_cli=None, versions=None):
|
|
"""Instantiate TestNode objects.
|
|
|
|
Should only be called once after the nodes have been specified in
|
|
set_test_params()."""
|
|
def get_bin_from_version(version, bin_name, bin_default):
|
|
if not version:
|
|
return bin_default
|
|
return os.path.join(
|
|
self.options.previous_releases_path,
|
|
re.sub(
|
|
r'\.0$',
|
|
'', # remove trailing .0 for point releases
|
|
'v{}.{}.{}.{}'.format(
|
|
(version % 100000000) // 1000000,
|
|
(version % 1000000) // 10000,
|
|
(version % 10000) // 100,
|
|
(version % 100) // 1,
|
|
),
|
|
),
|
|
'bin',
|
|
bin_name,
|
|
)
|
|
|
|
if self.bind_to_localhost_only:
|
|
extra_confs = [["bind=127.0.0.1"]] * num_nodes
|
|
else:
|
|
extra_confs = [[]] * num_nodes
|
|
if extra_args is None:
|
|
extra_args = [[]] * num_nodes
|
|
if versions is None:
|
|
versions = [None] * num_nodes
|
|
if self.is_syscall_sandbox_compiled() and not self.disable_syscall_sandbox:
|
|
for i in range(len(extra_args)):
|
|
if versions[i] is None or versions[i] >= 219900:
|
|
extra_args[i] = extra_args[i] + ["-sandbox=log-and-abort"]
|
|
if binary is None:
|
|
binary = [get_bin_from_version(v, 'bitcoind', self.options.bitcoind) for v in versions]
|
|
if binary_cli is None:
|
|
binary_cli = [get_bin_from_version(v, 'bitcoin-cli', self.options.bitcoincli) for v in versions]
|
|
assert_equal(len(extra_confs), num_nodes)
|
|
assert_equal(len(extra_args), num_nodes)
|
|
assert_equal(len(versions), num_nodes)
|
|
assert_equal(len(binary), num_nodes)
|
|
assert_equal(len(binary_cli), num_nodes)
|
|
for i in range(num_nodes):
|
|
test_node_i = TestNode(
|
|
i,
|
|
get_datadir_path(self.options.tmpdir, i),
|
|
chain=self.chain,
|
|
rpchost=rpchost,
|
|
timewait=self.rpc_timeout,
|
|
timeout_factor=self.options.timeout_factor,
|
|
bitcoind=binary[i],
|
|
bitcoin_cli=binary_cli[i],
|
|
version=versions[i],
|
|
coverage_dir=self.options.coveragedir,
|
|
cwd=self.options.tmpdir,
|
|
extra_conf=extra_confs[i],
|
|
extra_args=extra_args[i],
|
|
use_cli=self.options.usecli,
|
|
start_perf=self.options.perf,
|
|
use_valgrind=self.options.valgrind,
|
|
descriptors=self.options.descriptors,
|
|
)
|
|
self.nodes.append(test_node_i)
|
|
if not test_node_i.version_is_at_least(170000):
|
|
# adjust conf for pre 17
|
|
conf_file = test_node_i.bitcoinconf
|
|
with open(conf_file, 'r', encoding='utf8') as conf:
|
|
conf_data = conf.read()
|
|
with open(conf_file, 'w', encoding='utf8') as conf:
|
|
conf.write(conf_data.replace('[regtest]', ''))
|
|
|
|
def start_node(self, i, *args, **kwargs):
|
|
"""Start a bitcoind"""
|
|
|
|
node = self.nodes[i]
|
|
|
|
node.start(*args, **kwargs)
|
|
node.wait_for_rpc_connection()
|
|
|
|
if self.options.coveragedir is not None:
|
|
coverage.write_all_rpc_commands(self.options.coveragedir, node.rpc)
|
|
|
|
def start_nodes(self, extra_args=None, *args, **kwargs):
|
|
"""Start multiple bitcoinds"""
|
|
|
|
if extra_args is None:
|
|
extra_args = [None] * self.num_nodes
|
|
assert_equal(len(extra_args), self.num_nodes)
|
|
try:
|
|
for i, node in enumerate(self.nodes):
|
|
node.start(extra_args[i], *args, **kwargs)
|
|
for node in self.nodes:
|
|
node.wait_for_rpc_connection()
|
|
except:
|
|
# If one node failed to start, stop the others
|
|
self.stop_nodes()
|
|
raise
|
|
|
|
if self.options.coveragedir is not None:
|
|
for node in self.nodes:
|
|
coverage.write_all_rpc_commands(self.options.coveragedir, node.rpc)
|
|
|
|
def stop_node(self, i, expected_stderr='', wait=0):
|
|
"""Stop a bitcoind test node"""
|
|
self.nodes[i].stop_node(expected_stderr, wait=wait)
|
|
|
|
def stop_nodes(self, wait=0):
|
|
"""Stop multiple bitcoind test nodes"""
|
|
for node in self.nodes:
|
|
# Issue RPC to stop nodes
|
|
node.stop_node(wait=wait, wait_until_stopped=False)
|
|
|
|
for node in self.nodes:
|
|
# Wait for nodes to stop
|
|
node.wait_until_stopped()
|
|
|
|
def restart_node(self, i, extra_args=None):
|
|
"""Stop and start a test node"""
|
|
self.stop_node(i)
|
|
self.start_node(i, extra_args)
|
|
|
|
def wait_for_node_exit(self, i, timeout):
|
|
self.nodes[i].process.wait(timeout)
|
|
|
|
def connect_nodes(self, a, b):
|
|
from_connection = self.nodes[a]
|
|
to_connection = self.nodes[b]
|
|
ip_port = "127.0.0.1:" + str(p2p_port(b))
|
|
from_connection.addnode(ip_port, "onetry")
|
|
# poll until version handshake complete to avoid race conditions
|
|
# with transaction relaying
|
|
# See comments in net_processing:
|
|
# * Must have a version message before anything else
|
|
# * Must have a verack message before anything else
|
|
wait_until_helper(lambda: all(peer['version'] != 0 for peer in from_connection.getpeerinfo()))
|
|
wait_until_helper(lambda: all(peer['version'] != 0 for peer in to_connection.getpeerinfo()))
|
|
wait_until_helper(lambda: all(peer['bytesrecv_per_msg'].pop('verack', 0) == 24 for peer in from_connection.getpeerinfo()))
|
|
wait_until_helper(lambda: all(peer['bytesrecv_per_msg'].pop('verack', 0) == 24 for peer in to_connection.getpeerinfo()))
|
|
|
|
def disconnect_nodes(self, a, b):
|
|
def disconnect_nodes_helper(from_connection, node_num):
|
|
def get_peer_ids():
|
|
result = []
|
|
for peer in from_connection.getpeerinfo():
|
|
if "testnode{}".format(node_num) in peer['subver']:
|
|
result.append(peer['id'])
|
|
return result
|
|
|
|
peer_ids = get_peer_ids()
|
|
if not peer_ids:
|
|
self.log.warning("disconnect_nodes: {} and {} were not connected".format(
|
|
from_connection.index,
|
|
node_num,
|
|
))
|
|
return
|
|
for peer_id in peer_ids:
|
|
try:
|
|
from_connection.disconnectnode(nodeid=peer_id)
|
|
except JSONRPCException as e:
|
|
# If this node is disconnected between calculating the peer id
|
|
# and issuing the disconnect, don't worry about it.
|
|
# This avoids a race condition if we're mass-disconnecting peers.
|
|
if e.error['code'] != -29: # RPC_CLIENT_NODE_NOT_CONNECTED
|
|
raise
|
|
|
|
# wait to disconnect
|
|
wait_until_helper(lambda: not get_peer_ids(), timeout=5)
|
|
|
|
disconnect_nodes_helper(self.nodes[a], b)
|
|
|
|
def split_network(self):
|
|
"""
|
|
Split the network of four nodes into nodes 0/1 and 2/3.
|
|
"""
|
|
self.disconnect_nodes(1, 2)
|
|
self.sync_all(self.nodes[:2])
|
|
self.sync_all(self.nodes[2:])
|
|
|
|
def join_network(self):
|
|
"""
|
|
Join the (previously split) network halves together.
|
|
"""
|
|
self.connect_nodes(1, 2)
|
|
self.sync_all()
|
|
|
|
def no_op(self):
|
|
pass
|
|
|
|
def generate(self, generator, *args, sync_fun=None, **kwargs):
|
|
blocks = generator.generate(*args, invalid_call=False, **kwargs)
|
|
sync_fun() if sync_fun else self.sync_all()
|
|
return blocks
|
|
|
|
def generateblock(self, generator, *args, sync_fun=None, **kwargs):
|
|
blocks = generator.generateblock(*args, invalid_call=False, **kwargs)
|
|
sync_fun() if sync_fun else self.sync_all()
|
|
return blocks
|
|
|
|
def generatetoaddress(self, generator, *args, sync_fun=None, **kwargs):
|
|
blocks = generator.generatetoaddress(*args, invalid_call=False, **kwargs)
|
|
sync_fun() if sync_fun else self.sync_all()
|
|
return blocks
|
|
|
|
def generatetodescriptor(self, generator, *args, sync_fun=None, **kwargs):
|
|
blocks = generator.generatetodescriptor(*args, invalid_call=False, **kwargs)
|
|
sync_fun() if sync_fun else self.sync_all()
|
|
return blocks
|
|
|
|
def sync_blocks(self, nodes=None, wait=1, timeout=60):
|
|
"""
|
|
Wait until everybody has the same tip.
|
|
sync_blocks needs to be called with an rpc_connections set that has least
|
|
one node already synced to the latest, stable tip, otherwise there's a
|
|
chance it might return before all nodes are stably synced.
|
|
"""
|
|
rpc_connections = nodes or self.nodes
|
|
timeout = int(timeout * self.options.timeout_factor)
|
|
stop_time = time.time() + timeout
|
|
while time.time() <= stop_time:
|
|
best_hash = [x.getbestblockhash() for x in rpc_connections]
|
|
if best_hash.count(best_hash[0]) == len(rpc_connections):
|
|
return
|
|
# Check that each peer has at least one connection
|
|
assert (all([len(x.getpeerinfo()) for x in rpc_connections]))
|
|
time.sleep(wait)
|
|
raise AssertionError("Block sync timed out after {}s:{}".format(
|
|
timeout,
|
|
"".join("\n {!r}".format(b) for b in best_hash),
|
|
))
|
|
|
|
def sync_mempools(self, nodes=None, wait=1, timeout=60, flush_scheduler=True):
|
|
"""
|
|
Wait until everybody has the same transactions in their memory
|
|
pools
|
|
"""
|
|
rpc_connections = nodes or self.nodes
|
|
timeout = int(timeout * self.options.timeout_factor)
|
|
stop_time = time.time() + timeout
|
|
while time.time() <= stop_time:
|
|
pool = [set(r.getrawmempool()) for r in rpc_connections]
|
|
if pool.count(pool[0]) == len(rpc_connections):
|
|
if flush_scheduler:
|
|
for r in rpc_connections:
|
|
r.syncwithvalidationinterfacequeue()
|
|
return
|
|
# Check that each peer has at least one connection
|
|
assert (all([len(x.getpeerinfo()) for x in rpc_connections]))
|
|
time.sleep(wait)
|
|
raise AssertionError("Mempool sync timed out after {}s:{}".format(
|
|
timeout,
|
|
"".join("\n {!r}".format(m) for m in pool),
|
|
))
|
|
|
|
def sync_all(self, nodes=None):
|
|
self.sync_blocks(nodes)
|
|
self.sync_mempools(nodes)
|
|
|
|
def wait_until(self, test_function, timeout=60):
|
|
return wait_until_helper(test_function, timeout=timeout, timeout_factor=self.options.timeout_factor)
|
|
|
|
# Private helper methods. These should not be accessed by the subclass test scripts.
|
|
|
|
def _start_logging(self):
|
|
# Add logger and logging handlers
|
|
self.log = logging.getLogger('TestFramework')
|
|
self.log.setLevel(logging.DEBUG)
|
|
# Create file handler to log all messages
|
|
fh = logging.FileHandler(self.options.tmpdir + '/test_framework.log', encoding='utf-8')
|
|
fh.setLevel(logging.DEBUG)
|
|
# Create console handler to log messages to stderr. By default this logs only error messages, but can be configured with --loglevel.
|
|
ch = logging.StreamHandler(sys.stdout)
|
|
# User can provide log level as a number or string (eg DEBUG). loglevel was caught as a string, so try to convert it to an int
|
|
ll = int(self.options.loglevel) if self.options.loglevel.isdigit() else self.options.loglevel.upper()
|
|
ch.setLevel(ll)
|
|
# Format logs the same as bitcoind's debug.log with microprecision (so log files can be concatenated and sorted)
|
|
formatter = logging.Formatter(fmt='%(asctime)s.%(msecs)03d000Z %(name)s (%(levelname)s): %(message)s', datefmt='%Y-%m-%dT%H:%M:%S')
|
|
formatter.converter = time.gmtime
|
|
fh.setFormatter(formatter)
|
|
ch.setFormatter(formatter)
|
|
# add the handlers to the logger
|
|
self.log.addHandler(fh)
|
|
self.log.addHandler(ch)
|
|
|
|
if self.options.trace_rpc:
|
|
rpc_logger = logging.getLogger("BitcoinRPC")
|
|
rpc_logger.setLevel(logging.DEBUG)
|
|
rpc_handler = logging.StreamHandler(sys.stdout)
|
|
rpc_handler.setLevel(logging.DEBUG)
|
|
rpc_logger.addHandler(rpc_handler)
|
|
|
|
def _initialize_chain(self):
|
|
"""Initialize a pre-mined blockchain for use by the test.
|
|
|
|
Create a cache of a 199-block-long chain
|
|
Afterward, create num_nodes copies from the cache."""
|
|
|
|
CACHE_NODE_ID = 0 # Use node 0 to create the cache for all other nodes
|
|
cache_node_dir = get_datadir_path(self.options.cachedir, CACHE_NODE_ID)
|
|
assert self.num_nodes <= MAX_NODES
|
|
|
|
if not os.path.isdir(cache_node_dir):
|
|
self.log.debug("Creating cache directory {}".format(cache_node_dir))
|
|
|
|
initialize_datadir(self.options.cachedir, CACHE_NODE_ID, self.chain, self.disable_autoconnect)
|
|
self.nodes.append(
|
|
TestNode(
|
|
CACHE_NODE_ID,
|
|
cache_node_dir,
|
|
chain=self.chain,
|
|
extra_conf=["bind=127.0.0.1"],
|
|
extra_args=['-disablewallet'],
|
|
rpchost=None,
|
|
timewait=self.rpc_timeout,
|
|
timeout_factor=self.options.timeout_factor,
|
|
bitcoind=self.options.bitcoind,
|
|
bitcoin_cli=self.options.bitcoincli,
|
|
coverage_dir=None,
|
|
cwd=self.options.tmpdir,
|
|
descriptors=self.options.descriptors,
|
|
))
|
|
self.start_node(CACHE_NODE_ID)
|
|
cache_node = self.nodes[CACHE_NODE_ID]
|
|
|
|
# Wait for RPC connections to be ready
|
|
cache_node.wait_for_rpc_connection()
|
|
|
|
# Set a time in the past, so that blocks don't end up in the future
|
|
cache_node.setmocktime(cache_node.getblockheader(cache_node.getbestblockhash())['time'])
|
|
|
|
# Create a 199-block-long chain; each of the 3 first nodes
|
|
# gets 25 mature blocks and 25 immature.
|
|
# The 4th address gets 25 mature and only 24 immature blocks so that the very last
|
|
# block in the cache does not age too much (have an old tip age).
|
|
# This is needed so that we are out of IBD when the test starts,
|
|
# see the tip age check in IsInitialBlockDownload().
|
|
gen_addresses = [k.address for k in TestNode.PRIV_KEYS][:3] + [create_deterministic_address_bcrt1_p2tr_op_true()[0]]
|
|
assert_equal(len(gen_addresses), 4)
|
|
for i in range(8):
|
|
self.generatetoaddress(
|
|
cache_node,
|
|
nblocks=25 if i != 7 else 24,
|
|
address=gen_addresses[i % len(gen_addresses)],
|
|
)
|
|
|
|
assert_equal(cache_node.getblockchaininfo()["blocks"], 199)
|
|
|
|
# Shut it down, and clean up cache directories:
|
|
self.stop_nodes()
|
|
self.nodes = []
|
|
|
|
def cache_path(*paths):
|
|
return os.path.join(cache_node_dir, self.chain, *paths)
|
|
|
|
os.rmdir(cache_path('wallets')) # Remove empty wallets dir
|
|
for entry in os.listdir(cache_path()):
|
|
if entry not in ['chainstate', 'blocks', 'indexes']: # Only indexes, chainstate and blocks folders
|
|
os.remove(cache_path(entry))
|
|
|
|
for i in range(self.num_nodes):
|
|
self.log.debug("Copy cache directory {} to node {}".format(cache_node_dir, i))
|
|
to_dir = get_datadir_path(self.options.tmpdir, i)
|
|
shutil.copytree(cache_node_dir, to_dir)
|
|
initialize_datadir(self.options.tmpdir, i, self.chain, self.disable_autoconnect) # Overwrite port/rpcport in bitcoin.conf
|
|
|
|
def _initialize_chain_clean(self):
|
|
"""Initialize empty blockchain for use by the test.
|
|
|
|
Create an empty blockchain and num_nodes wallets.
|
|
Useful if a test case wants complete control over initialization."""
|
|
for i in range(self.num_nodes):
|
|
initialize_datadir(self.options.tmpdir, i, self.chain, self.disable_autoconnect)
|
|
|
|
def skip_if_no_py3_zmq(self):
|
|
"""Attempt to import the zmq package and skip the test if the import fails."""
|
|
try:
|
|
import zmq # noqa
|
|
except ImportError:
|
|
raise SkipTest("python3-zmq module not available.")
|
|
|
|
def skip_if_no_bitcoind_zmq(self):
|
|
"""Skip the running test if bitcoind has not been compiled with zmq support."""
|
|
if not self.is_zmq_compiled():
|
|
raise SkipTest("bitcoind has not been built with zmq enabled.")
|
|
|
|
def skip_if_no_wallet(self):
|
|
"""Skip the running test if wallet has not been compiled."""
|
|
self.requires_wallet = True
|
|
if not self.is_wallet_compiled():
|
|
raise SkipTest("wallet has not been compiled.")
|
|
if self.options.descriptors:
|
|
self.skip_if_no_sqlite()
|
|
else:
|
|
self.skip_if_no_bdb()
|
|
|
|
def skip_if_no_sqlite(self):
|
|
"""Skip the running test if sqlite has not been compiled."""
|
|
if not self.is_sqlite_compiled():
|
|
raise SkipTest("sqlite has not been compiled.")
|
|
|
|
def skip_if_no_bdb(self):
|
|
"""Skip the running test if BDB has not been compiled."""
|
|
if not self.is_bdb_compiled():
|
|
raise SkipTest("BDB has not been compiled.")
|
|
|
|
def skip_if_no_wallet_tool(self):
|
|
"""Skip the running test if bitcoin-wallet has not been compiled."""
|
|
if not self.is_wallet_tool_compiled():
|
|
raise SkipTest("bitcoin-wallet has not been compiled")
|
|
|
|
def skip_if_no_cli(self):
|
|
"""Skip the running test if bitcoin-cli has not been compiled."""
|
|
if not self.is_cli_compiled():
|
|
raise SkipTest("bitcoin-cli has not been compiled.")
|
|
|
|
def skip_if_no_previous_releases(self):
|
|
"""Skip the running test if previous releases are not available."""
|
|
if not self.has_previous_releases():
|
|
raise SkipTest("previous releases not available or disabled")
|
|
|
|
def has_previous_releases(self):
|
|
"""Checks whether previous releases are present and enabled."""
|
|
if not os.path.isdir(self.options.previous_releases_path):
|
|
if self.options.prev_releases:
|
|
raise AssertionError("Force test of previous releases but releases missing: {}".format(
|
|
self.options.previous_releases_path))
|
|
return self.options.prev_releases
|
|
|
|
def skip_if_no_external_signer(self):
|
|
"""Skip the running test if external signer support has not been compiled."""
|
|
if not self.is_external_signer_compiled():
|
|
raise SkipTest("external signer support has not been compiled.")
|
|
|
|
def is_cli_compiled(self):
|
|
"""Checks whether bitcoin-cli was compiled."""
|
|
return self.config["components"].getboolean("ENABLE_CLI")
|
|
|
|
def is_external_signer_compiled(self):
|
|
"""Checks whether external signer support was compiled."""
|
|
return self.config["components"].getboolean("ENABLE_EXTERNAL_SIGNER")
|
|
|
|
def is_wallet_compiled(self):
|
|
"""Checks whether the wallet module was compiled."""
|
|
return self.config["components"].getboolean("ENABLE_WALLET")
|
|
|
|
def is_specified_wallet_compiled(self):
|
|
"""Checks whether wallet support for the specified type
|
|
(legacy or descriptor wallet) was compiled."""
|
|
if self.options.descriptors:
|
|
return self.is_sqlite_compiled()
|
|
else:
|
|
return self.is_bdb_compiled()
|
|
|
|
def is_wallet_tool_compiled(self):
|
|
"""Checks whether bitcoin-wallet was compiled."""
|
|
return self.config["components"].getboolean("ENABLE_WALLET_TOOL")
|
|
|
|
def is_zmq_compiled(self):
|
|
"""Checks whether the zmq module was compiled."""
|
|
return self.config["components"].getboolean("ENABLE_ZMQ")
|
|
|
|
def is_sqlite_compiled(self):
|
|
"""Checks whether the wallet module was compiled with Sqlite support."""
|
|
return self.config["components"].getboolean("USE_SQLITE")
|
|
|
|
def is_bdb_compiled(self):
|
|
"""Checks whether the wallet module was compiled with BDB support."""
|
|
return self.config["components"].getboolean("USE_BDB")
|
|
|
|
def is_syscall_sandbox_compiled(self):
|
|
"""Checks whether the syscall sandbox was compiled."""
|
|
return self.config["components"].getboolean("ENABLE_SYSCALL_SANDBOX")
|