[version3] Deadlock on node closing #417
Description
(Posted here by Eric indication).
Hello.
I am experimenting a deadlock when I try to close the node (bn) Ctrl-C (SIGINT, SIGTERM).
To get the bn executable I follow:
wget https://raw.githubusercontent.com/libbitcoin/libbitcoin-node/version3/install.sh
chmod +x install.sh
sudo ./install.sh --prefix=/opt/libbitcoin --build-boost --disable-shared
#Boost modified to use 1.64 because 1.57 has compilation errors (multiprecision)
Using the following configuration file:
# cat /home/ubuntu/fer/deadlock/data/deadlock-btc.cfg
[log]
archive_directory = /home/ubuntu/fer/deadlock/data/log/archive-node-btc-testnet
debug_file = /home/ubuntu/fer/deadlock/data/log/node-btc-testnet-debug.log
error_file = /home/ubuntu/fer/deadlock/data/log/node-btc-testnet-error.log
rotation_size = 100000000
minimum_free_space = 0
verbose = true
[network]
protocol_maximum = 70013
protocol_minimum = 31402
identifier = 118034699
outbound_connections = 50
inbound_connections = 50
inbound_port = 18333
channel_expiration_minutes = 30
hosts_file = /home/ubuntu/fer/deadlock/data/hosts-btc-testnet.cache
seed = testnet-seed.bitcoin.jonasschnelli.ch:18333
seed = seed.tbtc.petertodd.org:18333
seed = testnet-seed.bluematt.me:18333
seed = testnet-seed.bitcoin.schildbach.de:18333
seed = testnet-seed.voskuil.org:18333
self = 0.0.0.0:0
[database]
directory = /home/ubuntu/fer/deadlock/data/database/btc-testnet
flush_writes = false
[blockchain]
# cores = 4
# Testnet-btc
checkpoint = 000000000933ea01ad0ee984209779baaec3ced90fa3f408719526f8d77f4943:0
checkpoint = 00000000009e2958c15ff9290d571bf9459e93b19765c6801ddeccadbb160a1e:100000
checkpoint = 0000000000287bffd321963ef05feab753ebe274e1d78b2fd4e2bfe9ad3aa6f2:200000
checkpoint = 000000000000226f7618566e70a2b5e020e29579b46743f05348427239bf41a1:300000
checkpoint = 000000000598cbbb1e79057b79eef828c495d4fc31050e6b179c57d07d00367c:400000
checkpoint = 000000000001a7c0aaa2630fbb2c0e476aafffc60f82177375b2aaa22209f606:500000
checkpoint = 000000000000624f06c69d3a9fe8d25e0a9030569128d63ad1b704bbb3059a16:600000
checkpoint = 000000000000406178b12a4dea3b27e13b3c4fe4510994fd667d7c1e6a3f4dc1:700000
checkpoint = 0000000000209b091d6519187be7c2ee205293f25f9f503f90027e25abf8b503:800000
checkpoint = 0000000000356f8d8924556e765b7a94aaebc6b5c8685dcfa2b1ee8b41acd89b:900000
checkpoint = 0000000000478e259a3eda2fafbeeb0106626f946347955e99278fe6cc848414:1000000
checkpoint = 00000000001aa0b431dc7f8fa75179b8440bdb671db5ca79e1087faff00c19d8:1050000
checkpoint = 00000000001c2fb9880485b1f3d7b0ffa9fabdfd0cf16e29b122bb6275c73db0:1100000
[fork]
easy_blocks = true
# retarget = true
bip16 = true
bip30 = true
bip34 = true
bip66 = true
bip65 = true
bip90 = true
[node]
relay_transactions=true
... and I have the following Python script to run the node multiple times:
# cat multiple_launchs.py
from subprocess import call
from subprocess import check_output
from random import randint
from subprocess import Popen
from subprocess import CalledProcessError
import subprocess
import signal
import os
import time
def get_pid(name):
try:
return map(int, check_output(["pidof",name]).split())
except CalledProcessError:
return []
def process_is_running(name):
return len(get_pid(name)) > 0
def launch():
sout = open('sout-fer.log', 'w')
serr = open('serr-fer.log', 'w')
return (sout, serr, Popen(["./bn", "-c", "/home/ubuntu/fer/deadlock/data/deadlock-btc.cfg"], stdout=sout, stderr=serr))
def finalize(p):
pid = p.pid
os.kill(pid, signal.SIGINT)
def finalize_and_wait(p):
finalize(p)
return p.wait()
def finalize_wait_measure(p):
finalize(p)
start_time = time.time()
p.wait()
elapsed_time = time.time() - start_time
return elapsed_time
def wait_for_termination(name):
while process_is_running(name):
time.sleep(10)
def main():
stats = []
for i in range(0, 100):
sout, serr, p = launch()
seconds = randint(10, 120)
print("******************************* Sleeping for %s seconds..." % (seconds))
time.sleep(seconds)
print("******************************* SLEEP FINALIZED *******************************************")
elapsed_time = finalize_wait_measure(p)
sout.close()
serr.close()
print("******************************* Closing procedure elapsed time: %s seconds..." % (elapsed_time))
stats.append([i, seconds, elapsed_time])
print("******************************* STATS *******************************************")
print(stats)
main()
Node chain initialization and script running:
./bn -i -c /home/ubuntu/fer/deadlock/data/deadlock-btc.cfg
python multiple_launchs.py
(the python script have to be in the same directory as bn executable)
Thanks and regards,
Fernando.