deploy: persistent systemd unit running a pinned release binary with Restart=on-failure + boot enable, plus a liveness timer that restarts the daemon if a register round-trip stops answering

This commit is contained in:
Jean Chevronnet 2026-08-12 16:01:58 +00:00
parent b022189fa5
commit 2aaa5ac091
4 changed files with 76 additions and 0 deletions

View file

@ -0,0 +1,19 @@
[Unit]
Description=echoIRCd (production) — native Rust ircd
After=network-online.target
Wants=network-online.target
[Service]
Type=simple
User=debian
WorkingDirectory=/home/debian/irc/ircd/echoIRCd
LimitNOFILE=200000
# runs a PINNED release binary (bin/echoircd), not target/debug — so a stray
# `cargo build` never changes what a restart would launch. Deploy = copy a fresh
# `cargo build --release` binary over bin/echoircd, then `systemctl restart`.
ExecStart=/home/debian/irc/ircd/echoIRCd/bin/echoircd /home/debian/irc/ircd/echoIRCd/echoircd.conf
Restart=on-failure
RestartSec=2
[Install]
WantedBy=multi-user.target

View file

@ -0,0 +1,8 @@
[Unit]
Description=echoIRCd liveness probe — restart the daemon if it stops responding
[Service]
Type=oneshot
# a hung (but not exited) daemon isn't caught by Restart=on-failure; this probe
# does a real register round-trip and restarts the service if it doesn't answer.
ExecStart=/home/debian/irc/ircd/echoIRCd/scripts/liveness.sh 6767

View file

@ -0,0 +1,10 @@
[Unit]
Description=Run the echoIRCd liveness probe periodically
[Timer]
OnBootSec=2min
OnUnitActiveSec=2min
AccuracySec=15s
[Install]
WantedBy=timers.target

39
scripts/liveness.sh Executable file
View file

@ -0,0 +1,39 @@
#!/usr/bin/env bash
# echoIRCd liveness probe: do a real NICK/USER register round-trip on the plaintext
# port and, if the daemon doesn't answer with a 001 welcome, restart it. Catches a
# hung core (process alive but not serving), which Restart=on-failure alone can't.
# Run by echoircd-liveness.timer.
set -u
PORT="${1:-6767}"
if python3 - "$PORT" <<'PY'
import socket, sys, time
port = int(sys.argv[1])
try:
s = socket.create_connection(("127.0.0.1", port), timeout=5)
s.settimeout(5)
s.sendall(b"NICK livecheck\r\nUSER lc 0 * :liveness\r\n")
data = b""
end = time.time() + 6
while time.time() < end:
d = s.recv(4096)
if not d:
break
data += d
if b" 001 " in data:
break
try:
s.sendall(b"QUIT :liveness\r\n")
except OSError:
pass
s.close()
sys.exit(0 if b" 001 " in data else 1)
except OSError:
sys.exit(1)
PY
then
exit 0
fi
echo "echoircd liveness: no 001 from 127.0.0.1:${PORT} within timeout — restarting echoircd-dev.service"
systemctl restart echoircd-dev.service