- Remove 'uci: { tachyon: true }' dependency from menu.d/luci-app-tachyon.json so LuCI does not hide the menu node if UCI config is uncommitted/reloading
- Add portable timeout normalization in install.sh ucode to handle Busybox timeout and systems without timeout applet
- Ensure luci_postinst is called directly in IPK postinst (build.sh) and after install in install.sh
- Restart uhttpd alongside rpcd and purge LuCI caches in post-install and uninstaller
- Fix uninstall.sh to remove packages individually with lock retry and omit non-existent luci-i18n-tachyon-en
- In install_ui_packages, verify that menu.d/luci-app-tachyon.json actually exists on disk before proceeding
- action.uc: sweep stuck flock 1000 processes before tachyon restart, bound service restart with 25s timeout, and prevent worker lockup
- shell/index.ts: add 15-minute general hard timeout for non-Tachyon components to prevent infinite modal spin; allow same-version confirmation for both install and reinstall
- package.uc: remove disruptive uhttpd restart from luci_postinst to prevent active socket drops
- ui.uc: refresh stale PID state before checking value.running in action_ack so dead jobs can always be acknowledged
- initController.ts & sessionJobs.ts: persist handled job IDs across reloads in sessionStorage, eliminate redundant modal popups on page reload, and guard against concurrent reloads with safeReloadPage
- renderUpdateProgressModal.ts: record active job before reload and poll readiness with fast health check instead of blind 5s freeze
- Fix response parsing: check report field, not just message
- Pass user query text to LLM instead of ignoring it
- Append user question to diagnostic prompt with system context
- Increase timeout to 120s for LM Studio responses
- Add command_status_from_args to core/common.uc exports
Closes#38
- kill_matching_command и цикл зачистки fd/1000 прокси в package.uc теперь пропускют процессы ucode (action.uc / updates.uc) и их предков, предотвращая выбивание воркера обновлений по SIGKILL во время prerm
- isTransientRpcError теперь учитывает 'failed to execute' в качестве транзиентной ошибки RPC при перезапуске rpcd / служб во время обновления
Самообновление Tachyon — это компонентное задание, которое устанавливает
пакет, чей postinst и вызывает remove_component_update_cache(). Очистка
стирала каталог component-actions целиком, включая файл состояния того
самого задания, которое ещё выполнялось. LuCI опрашивал задание, не
находил его, не получал ни message, ни stderr и показывал фолбэк
"Failed to execute" поверх успешно прошедшего обновления.
Теперь очистка пропускает задания, помеченные running. Устаревшие
результаты и кэш проверок обновлений убираются как раньше. Зависшее
задание не залипнет: детект по PID и started_at помечает такие stale.
Пути каталогов состояния стали переопределяемыми через окружение, чтобы
тест мог прогнать luci-postinst, не трогая /var/run.
`ps | grep '/usr/bin/tachyon'` matches the hook itself, because
/usr/bin/tachyon is what invoked it. The sweep therefore killed its own
parent, the parent died on SIGKILL, and apk saw
`tachyon-<version>.pre-upgrade: exited with error 247` (128+119) and
reported a failed hook.
The sweep now walks /proc/<pid>/stat to collect its own ancestor chain and
skips those pids.
Holders of procd's serialization lock are also cleared by descriptor rather
than by name. procd_lock() does `exec 1000>` on
/var/lock/procd_tachyon.lock and then `flock 1000` with no -w, so any
background spawn inheriting fd 1000 keeps the lock held. On a live router
the holders were watchdog workers, logread, the zapret2 supervisors, nfqws2
and a stray curl — none of which match a tachyon name pattern. While any of
them lives, every init.d call blocks in flock until the package manager
gives up on the hook.
Same descriptor-based sweep added to install.sh, which faces the same lock
left behind by pre-fix builds.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The router upgrade died at the installer's 120s ceiling while apk was
running tachyon-1.2.66.pre-upgrade, so apk rolled the transaction back
and the old build stayed installed.
prerm reached INIT_PATH status through rc.common, which waits on
flock -w 1000. A stuck start or retry_start_on_wan_up holding that lock
stalls the hook for up to a quarter of an hour. Every init call in the
hooks is now bounded, the status probe moved after the flock waiters are
killed, and the Makefile hook bounds the old version's cleanup from the
outside since that code cannot be trusted to bound itself.
The installer also called a rollback a success: a rebuilt tag keeps its
version, so both the presence and version checks still passed. It now
compares the commit SHA baked into the installed constants module and
fails loudly when it did not move. Package installs get 300s, since a
kill mid-transaction is what caused the rollback.
- postinst_restore: switch from BIN_PATH start to INIT_PATH start so
the test-injected init script is exercised (the test sets TACHYON_INIT
to a fake init that writes to a log; BIN_PATH was never the fake init)
- postinst_restore: guard process-killing with !PACKAGE_TEST_MODE,
consistent with prerm_cleanup — the start call itself is always made
- flock deadlock risk is gone: prerm_cleanup now kills all waiters
before removing the routing table entry, so INIT_PATH start is safe
- service/package.uc: call INIT_PATH stop before process-killing in
prerm_cleanup so the routing table entry is still visible when the
service shuts down (fixes test: 'prerm must stop Tachyon before
removing its routing table name')
- service/package.uc: replace awk regex with | alternation and \/
escaping with grep -E / grep -F; BusyBox awk does not support |
in regex literals and silently mishandles \/ — fixes awk syntax
errors in CI (both prerm_cleanup and postinst_restore)
- fe-app-tachyon: pin js-yaml to ^5.2.1 via overrides to fix the
quadratic O(n^2) !!omap resolution DoS (same root cause as
CVE-2026-59870 / GHSA-724g-mxrg-4qvm, never backported to 3.x/4.x)
- fe-app-tachyon: npm audit fix for brace-expansion and nanoid vulns
Критические исправления:
- safe_proxy_restart: удаление lock-файла после фонового перезапуска
с защитой от stale lock (>300 сек)
- write_json_file: атомарная запись через tmp+mv вместо прямой
- write_reload_state: атомарная запись через tmp+mv
- state_add_list: отклонение значений с пробелами
Исправления безопасности:
- killall logread заменён на pkill -f 'logread -f'
- SIGTERM→SIGKILL для Telegram и sing-box при остановке
- Recover OOM scale: инкремент +0.05 каждые 10 минут
- nfqueue и byedpi: trap для корректного завершения дочерних процессов
- nfqueue: валидация стратегий и откат при ошибке запуска
- nfqueue: flush mangle_output перед добавлением правил
- Исправлен парсинг exit code в 14 файлах (status & 127)
Улучшения стабильности:
- init.d: trap для start_service/stop_service
- watchdog: signal traps для legacy fallback loop
- lifecycle: логирование ошибок при остановке компонентов
- singbox/priority: SIGTERM→SIGKILL при остановке
- telegram: exponential backoff (1x→300s) при ошибках API
- telegram: process_updates возвращает boolean для отслеживания успеха
- watchdog: кэш proxy port для уменьшения чтения конфига sing-box
- singbox/generator: упрощён atomic_write_json_file
- packages.uc: переупорядочивание apk_version (лёгкие запросы первыми)
- dns/apply: проверка PID sing-box перед настройкой dnsmasq
- server/service: cleanup_orphaned_server_certs перед вызовом
Dushnilin