The_Great_Archive/ArchiveBox
1
0
Fork 0
mirror of https://github.com/ArchiveBox/ArchiveBox.git synced 2026-09-12 19:50:57 +05:00
Code Issues Actions 4 Packages Projects Releases Wiki Activity

fix: centralize ctrl-c rich logging

Browse Source
This commit is contained in:
Nick Sweeting 2026-05-28 08:49:17 -07:00
parent 8cf42f9f65
commit f60b190182
No known key found for this signature in database
8 changed files with 141 additions and 32 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

32
archivebox/cli/__init__.py
View File

@ -5,10 +5,12 @@ import sys
from importlib import import_module
import rich_click as click
from rich import print
from rich.console import Console
from archivebox.config.version import VERSION
STDERR = Console(stderr=True)
if "--debug" in sys.argv:
os.environ["DEBUG"] = "True"
@ -104,9 +106,8 @@ class ArchiveBoxGroup(click.Group):
# handle renamed commands
if cmd_name in self.renamed_commands:
new_name = self.renamed_commands[cmd_name]
print(
STDERR.print(
f" [violet]Hint:[/violet] `archivebox {cmd_name}` has been renamed to `archivebox {new_name}`",
file=sys.stderr,
)
cmd_name = new_name
ctx.invoked_subcommand = cmd_name
@ -175,7 +176,7 @@ def cli(ctx, help=False):
if subcommand != "update":
check_migrations(auto_apply=True)
except Exception as e:
print(f"[red][X] Error setting up Django or checking data folder: {e}[/red]", file=sys.stderr)
STDERR.print(f"[red][X] Error setting up Django or checking data folder: {e}[/red]")
if subcommand not in ("manage", "shell"): # not all management commands need django to be setup beforehand
raise
@ -189,10 +190,29 @@ def main(args=None, prog_name=None, stdin=None):
# stdin param allows passing input data from caller (used by __main__.py)
# currently not used by click-based CLI, but kept for backwards compatibility
previous_unraisablehook = sys.unraisablehook
def ignore_shutdown_unraisable(unraisable):
if isinstance(unraisable.exc_value, (KeyboardInterrupt, SystemExit)):
return
previous_unraisablehook(unraisable)
sys.unraisablehook = ignore_shutdown_unraisable
try:
cli(args=args, prog_name=prog_name)
cli(args=args, prog_name=prog_name, standalone_mode=False)
except click.Abort:
STDERR.print("\n[red][X] Got CTRL+C. Exiting...[/red]")
raise SystemExit(130) from None
except click.ClickException as err:
err.show()
raise SystemExit(err.exit_code) from None
except click.exceptions.Exit as err:
raise SystemExit(err.exit_code) from None
except KeyboardInterrupt:
print("\n\n[red][X] Got CTRL+C. Exiting...[/red]")
STDERR.print("\n[red][X] Got CTRL+C. Exiting...[/red]")
raise SystemExit(130) from None
finally:
sys.unraisablehook = previous_unraisablehook
if __name__ == "__main__":

2
archivebox/config/django.py
View File

@ -162,6 +162,6 @@ def setup_django(check_db=False, in_memory_db=False) -> None:
# logfire.info(f'Started ArchiveBox v{CONSTANTS.VERSION}', argv=sys.argv)
except KeyboardInterrupt:
raise SystemExit(2)
raise
DJANGO_SET_UP = True

8
archivebox/core/models.py
View File

@ -27,7 +27,7 @@ from django.core.exceptions import ObjectDoesNotExist, ValidationError
from django.utils.safestring import mark_safe
from archivebox.config import CONSTANTS
from archivebox.config.common import get_config
from archivebox.config.common import get_config, rprint
from archivebox.misc.system import atomic_write
from archivebox.misc.util import (
MAX_URL_LENGTH,
@ -2207,7 +2207,7 @@ class Snapshot(ModelWithDeleteAfter, ModelWithOutputDir, ModelWithConfig, ModelW
deleted_count = empty_ars.count()
if deleted_count > 0:
empty_ars.delete()
print(f"[yellow]🗑️ Deleted {deleted_count} empty ArchiveResults for {self.url}[/yellow]")
rprint(f"[yellow]🗑️ Deleted {deleted_count} empty ArchiveResults for {self.url}[/yellow]")
def to_json(self) -> dict:
"""
@ -2315,7 +2315,7 @@ class Snapshot(ModelWithDeleteAfter, ModelWithOutputDir, ModelWithConfig, ModelW
record_crawl_id = record.get("crawl_id")
if record_crawl_id and crawl and str(crawl.id) != str(record_crawl_id):
print(
rprint(
f"[yellow]⚠️ Snapshot.from_json crawl mismatch: record has crawl_id={record_crawl_id}, overrides has crawl={crawl.id}[/yellow]",
file=sys.stderr,
)
@ -2341,7 +2341,7 @@ class Snapshot(ModelWithDeleteAfter, ModelWithOutputDir, ModelWithConfig, ModelW
label=f"auto-created for {url[:50]}",
created_by_id=created_by_id,
)
print(f"[red]⚠️ Snapshot.from_json auto-created new crawl {crawl.id} for url={url}[/red]", file=sys.stderr)
rprint(f"[red]⚠️ Snapshot.from_json auto-created new crawl {crawl.id} for url={url}[/red]", file=sys.stderr)
# Parse tags (accept either a list ["tag1", "tag2"] or a comma-separated string "tag1,tag2")
tags_raw = record.get("tags", "")

2
archivebox/crawls/models.py
View File

@ -21,7 +21,7 @@ from django.conf import settings
from django.urls import reverse_lazy
from django.utils import timezone
from statemachine import State, registry
from rich import print
from archivebox.config.common import rprint as print
from archivebox.base_models.models import (
ModelWithUUID,

13
archivebox/machine/models.py
View File

@ -19,6 +19,7 @@ from django.db.models import Q, QuerySet
from django.utils import timezone
from django.utils.functional import cached_property
from archivebox.config.common import rprint
from archivebox.base_models.models import ModelWithDeleteAfter, ModelWithHealthStats
from archivebox.workers.models import BaseStateMachine, ModelWithStateMachine
from .detect import get_host_guid, get_os_info, get_vm_info, get_host_network, get_host_stats
@ -2435,10 +2436,10 @@ class Process(ModelWithDeleteAfter, models.Model):
if result.returncode == 0:
killed += int(result.stdout.strip())
if killed > 0:
print(f"[yellow]🧹 Cleaned up {killed} orphaned Chrome processes[/yellow]")
rprint(f"[yellow]🧹 Cleaned up {killed} orphaned Chrome processes[/yellow]")
return killed
except (subprocess.TimeoutExpired, ValueError, FileNotFoundError) as e:
print(f"[red]Failed to cleanup orphaned Chrome: {e}[/red]")
rprint(f"[red]Failed to cleanup orphaned Chrome: {e}[/red]")
return 0
@ -2498,7 +2499,7 @@ class Process(ModelWithDeleteAfter, models.Model):
cleaned += 1
if cleaned:
print(f"[yellow]🧹 Cleaned up {cleaned} orphaned worker/hook process record(s)[/yellow]")
rprint(f"[yellow]🧹 Cleaned up {cleaned} orphaned worker/hook process record(s)[/yellow]")
return cleaned
@ -2553,7 +2554,7 @@ class BinaryMachine(BaseStateMachine):
"""Called during queued→installed transition. Runs installation synchronously."""
import sys
print(f"[cyan] 🔄 BinaryMachine.on_install() - installing {self.binary.name}[/cyan]", file=sys.stderr)
rprint(f"[cyan] 🔄 BinaryMachine.on_install() - installing {self.binary.name}[/cyan]", file=sys.stderr)
# Run installation hooks (synchronous, updates abspath/version/sha256 and sets status)
self.binary.run()
@ -2564,7 +2565,7 @@ class BinaryMachine(BaseStateMachine):
if self.binary.status != Binary.StatusChoices.INSTALLED:
# Installation failed - abort transition, stay in queued
print(f"[red] ❌ BinaryMachine - {self.binary.name} installation failed, retrying later[/red]", file=sys.stderr)
rprint(f"[red] ❌ BinaryMachine - {self.binary.name} installation failed, retrying later[/red]", file=sys.stderr)
# Bump retry_at to try again later
self.binary.update_and_requeue(
@ -2578,7 +2579,7 @@ class BinaryMachine(BaseStateMachine):
# Abort the transition - this will raise an exception and keep us in queued
raise Exception(f"Binary {self.binary.name} installation failed")
print(f"[cyan] ✅ BinaryMachine - {self.binary.name} installed successfully[/cyan]", file=sys.stderr)
rprint(f"[cyan] ✅ BinaryMachine - {self.binary.name} installed successfully[/cyan]", file=sys.stderr)
@installed.enter
def enter_installed(self):

8
archivebox/services/supervision_service.py
View File

@ -4,7 +4,7 @@ import time
from pathlib import Path
from django.utils import timezone
from rich import print
from archivebox.config.common import rprint
def runtime_stack_owner_types():
@ -109,7 +109,7 @@ def ensure_daemon_stack(*, reason: str = ""):
return worker
if reason:
print(f"[yellow][*] Starting daemon stack for {reason}...[/yellow]")
rprint(f"[yellow][*] Starting daemon stack for {reason}...[/yellow]")
return start_worker(supervisor, sonic_worker)
@ -143,7 +143,7 @@ def standby_until_leader_needed(command, *, process_type: str, data_dir: str | P
if not announced:
leader = newest_live_process(process_type=process_type, data_dir=data_dir, url=url)
leader_pid = leader.pid if leader else "unknown"
print(f"[yellow][*] Standing by; newer ArchiveBox parent pid={leader_pid} is running the orchestrator and server.[/yellow]")
rprint(f"[yellow][*] Standing by; newer ArchiveBox parent pid={leader_pid} is running the orchestrator and server.[/yellow]")
announced = True
command.heartbeat()
time.sleep(interval)
@ -161,7 +161,7 @@ def standby_until_runtime_stack_needed(command, *, data_dir: str | Path, interva
owner = runtime_stack_owner(data_dir=data_dir)
owner_pid = owner.pid if owner else "unknown"
owner_type = owner.process_type if owner else "unknown"
print(f"[yellow][*] Standing by; ArchiveBox {owner_type} pid={owner_pid} owns the runtime stack.[/yellow]")
rprint(f"[yellow][*] Standing by; ArchiveBox {owner_type} pid={owner_pid} owns the runtime stack.[/yellow]")
announced = True
command.heartbeat()
time.sleep(interval)

2
archivebox/workers/supervisord_util.py
View File

@ -13,11 +13,11 @@ from typing import cast
from pathlib import Path
from functools import cache
from rich import print
from supervisor.xmlrpc import SupervisorTransport
from xmlrpc.client import Fault, ServerProxy
from archivebox.config import CONSTANTS
from archivebox.config.common import rprint as print
from archivebox.config.paths import get_or_create_working_tmp_dir
from archivebox.config.permissions import ARCHIVEBOX_USER
from archivebox.core.shutdown_util import (

106
bin/fuzz_test.sh
View File

@ -12,6 +12,10 @@ FUZZ_ROUNDS="${FUZZ_ROUNDS:-8}"
FUZZ_PARALLEL="${FUZZ_PARALLEL:-5}"
FUZZ_KILL_MIN_SECONDS="${FUZZ_KILL_MIN_SECONDS:-60}"
FUZZ_KILL_MAX_SECONDS="${FUZZ_KILL_MAX_SECONDS:-120}"
FUZZ_CTRL_C_CHANCE="${FUZZ_CTRL_C_CHANCE:-60}"
FUZZ_CTRL_C_MAX_SIGNALS="${FUZZ_CTRL_C_MAX_SIGNALS:-4}"
FUZZ_CTRL_C_MIN_SECONDS="${FUZZ_CTRL_C_MIN_SECONDS:-2}"
FUZZ_CTRL_C_MAX_SECONDS="${FUZZ_CTRL_C_MAX_SECONDS:-20}"
SERVER_BASE_PORT="${FUZZ_SERVER_BASE_PORT:-8700}"
SLEEP_BETWEEN_JOBS_MAX="${FUZZ_SLEEP_BETWEEN_JOBS_MAX:-5}"
@ -21,6 +25,12 @@ fi
if [[ "$FUZZ_KILL_MAX_SECONDS" -lt "$FUZZ_KILL_MIN_SECONDS" ]]; then
FUZZ_KILL_MAX_SECONDS="$FUZZ_KILL_MIN_SECONDS"
fi
if [[ "$FUZZ_CTRL_C_MAX_SECONDS" -lt "$FUZZ_CTRL_C_MIN_SECONDS" ]]; then
FUZZ_CTRL_C_MAX_SECONDS="$FUZZ_CTRL_C_MIN_SECONDS"
fi
if [[ "$FUZZ_CTRL_C_MAX_SIGNALS" -lt 1 ]]; then
FUZZ_CTRL_C_MAX_SIGNALS=1
fi
if [[ -n "${ARCHIVEBOX_CMD:-}" ]]; then
read -r -a ABX <<< "$ARCHIVEBOX_CMD"
@ -73,15 +83,76 @@ random_start_delay() {
random_between 0 "$SLEEP_BETWEEN_JOBS_MAX"
}
kill_tree() {
local pid="$1"
random_ctrl_c_delay() {
random_between "$FUZZ_CTRL_C_MIN_SECONDS" "$FUZZ_CTRL_C_MAX_SECONDS"
}
random_subsecond_delay() {
printf '0.%03d\n' "$((100 + RANDOM % 400))"
}
signal_tree() {
local signal="$1"
local pid="$2"
local child
if command -v pgrep >/dev/null 2>&1; then
for child in $(pgrep -P "$pid" 2>/dev/null || true); do
kill_tree "$child"
signal_tree "$signal" "$child"
done
fi
kill "$pid" >/dev/null 2>&1 || true
kill "-$signal" "$pid" >/dev/null 2>&1 || true
}
is_uv_wrapper_without_child() {
local pid="$1"
local comm
comm="$(ps -p "$pid" -o comm= 2>/dev/null | xargs basename 2>/dev/null || true)"
[[ "$comm" == "uv" ]] && ! pgrep -P "$pid" >/dev/null 2>&1
}
kill_tree() {
local pid="$1"
signal_tree TERM "$pid"
}
start_ctrl_c_injector() {
local label="$1"
local child="$2"
local logfile="$3"
local signals idx delay burst_gap
if [[ "$FUZZ_CTRL_C_CHANCE" -le 0 || $((RANDOM % 100)) -ge "$FUZZ_CTRL_C_CHANCE" ]]; then
return 0
fi
signals=$((1 + RANDOM % FUZZ_CTRL_C_MAX_SIGNALS))
(
for idx in $(seq 1 "$signals"); do
if [[ "$idx" -eq 1 || $((RANDOM % 2)) -eq 0 ]]; then
delay="$(random_ctrl_c_delay)"
else
delay="$(random_subsecond_delay)"
fi
sleep "$delay"
if ! kill -0 "$child" >/dev/null 2>&1; then
exit 0
fi
if is_uv_wrapper_without_child "$child"; then
exit 0
fi
echo "[$(ts)] CTRL_C label=$label pid=$child signal=$idx/$signals delay=${delay}s" >> "$logfile"
signal_tree INT "$child"
if [[ $((RANDOM % 3)) -eq 0 ]]; then
burst_gap="$(random_subsecond_delay)"
sleep "$burst_gap"
if kill -0 "$child" >/dev/null 2>&1 && ! is_uv_wrapper_without_child "$child"; then
echo "[$(ts)] CTRL_C_BURST label=$label pid=$child gap=${burst_gap}s" >> "$logfile"
signal_tree INT "$child"
fi
fi
done
) &
echo "$!"
}
cleanup() {
@ -112,30 +183,39 @@ run_with_timeout() {
echo "[$(ts)] START label=$label shell=$$ data=$DATA_DIR"
echo "[$(ts)] CMD DATA_DIR=$DATA_DIR $*"
echo "[$(ts)] CHAOS kill_after=${timeout}s"
echo "[$(ts)] CTRL_C chance=${FUZZ_CTRL_C_CHANCE}% max_signals=${FUZZ_CTRL_C_MAX_SIGNALS}"
} | tee -a "$logfile"
(
DATA_DIR="$DATA_DIR" "$@"
) >> "$logfile" 2>&1 &
local child=$!
local interrupter
interrupter="$(start_ctrl_c_injector "$label" "$child" "$logfile")"
(
sleep "$timeout"
if kill -0 "$child" >/dev/null 2>&1; then
echo "[$(ts)] TIMEOUT label=$label pid=$child after=${timeout}s" >> "$logfile"
kill "$child" >/dev/null 2>&1 || true
signal_tree TERM "$child"
sleep 5
kill -9 "$child" >/dev/null 2>&1 || true
signal_tree KILL "$child"
fi
) &
local watchdog=$!
trap '[[ -n "${child:-}" ]] && kill_tree "$child"; [[ -n "${watchdog:-}" ]] && kill "$watchdog" >/dev/null 2>&1 || true' INT TERM
trap '[[ -n "${child:-}" ]] && kill_tree "$child"; [[ -n "${watchdog:-}" ]] && kill "$watchdog" >/dev/null 2>&1 || true; [[ -n "${interrupter:-}" ]] && kill "$interrupter" >/dev/null 2>&1 || true' INT TERM
wait "$child"
local code=$?
kill "$watchdog" >/dev/null 2>&1 || true
if [[ -n "$interrupter" ]]; then
kill "$interrupter" >/dev/null 2>&1 || true
fi
wait "$watchdog" >/dev/null 2>&1 || true
if [[ -n "$interrupter" ]]; then
wait "$interrupter" >/dev/null 2>&1 || true
fi
trap - INT TERM
echo "[$(ts)] END label=$label pid=$child exit=$code log=$logfile" | tee -a "$logfile"
@ -157,22 +237,29 @@ run_server_for_a_bit() {
echo "[$(ts)] START label=$label shell=$$ data=$DATA_DIR"
echo "[$(ts)] CMD DATA_DIR=$DATA_DIR ${ABX[*]} server $debug_flag 127.0.0.1:$port"
echo "[$(ts)] CHAOS kill_after=${hold}s"
echo "[$(ts)] CTRL_C chance=${FUZZ_CTRL_C_CHANCE}% max_signals=${FUZZ_CTRL_C_MAX_SIGNALS}"
} | tee -a "$logfile"
(
DATA_DIR="$DATA_DIR" "${ABX[@]}" server "${server_extra[@]}" "127.0.0.1:$port"
) >> "$logfile" 2>&1 &
local child=$!
local interrupter
interrupter="$(start_ctrl_c_injector "$label" "$child" "$logfile")"
trap '[[ -n "${child:-}" ]] && kill_tree "$child"' INT TERM
trap '[[ -n "${child:-}" ]] && kill_tree "$child"; [[ -n "${interrupter:-}" ]] && kill "$interrupter" >/dev/null 2>&1 || true' INT TERM
sleep "$hold"
echo "[$(ts)] STOP label=$label pid=$child after=${hold}s" | tee -a "$logfile"
kill_tree "$child"
sleep 5
kill -9 "$child" >/dev/null 2>&1 || true
signal_tree KILL "$child"
wait "$child" >/dev/null 2>&1
local code=$?
if [[ -n "$interrupter" ]]; then
kill "$interrupter" >/dev/null 2>&1 || true
wait "$interrupter" >/dev/null 2>&1 || true
fi
trap - INT TERM
echo "[$(ts)] END label=$label pid=$child exit=$code log=$logfile" | tee -a "$logfile"
@ -298,6 +385,7 @@ main() {
echo " rounds: $FUZZ_ROUNDS"
echo " parallel: $FUZZ_PARALLEL"
echo " kill after: ${FUZZ_KILL_MIN_SECONDS}s-${FUZZ_KILL_MAX_SECONDS}s"
echo " ctrl+c: ${FUZZ_CTRL_C_CHANCE}% chance, ${FUZZ_CTRL_C_MAX_SIGNALS} max, ${FUZZ_CTRL_C_MIN_SECONDS}s-${FUZZ_CTRL_C_MAX_SECONDS}s plus bursts"
echo " urls: ${URLS[*]}"
echo