peter
6b0401b922
114 exception handlers neither logged nor re-raised. The defect was never that
they are silent — plenty of them should be — it is that their silence said
nothing, so nobody could tell a decision from an oversight. Each one has now been
read. 44 got a line; the other 70 carry `# silent-ok: <why>`.
What the silent ones had been hiding, in the cases that got a line:
- A failed ffprobe sent every HDR file down the 8-bit pipeline, so playback
degraded to a software transcode with nothing saying why; a failed audio probe
collapsed a multi-audio file to one track and the player showed no language
picker; a failed first-PTS probe left subtitles seconds out of sync.
- A subtitle sidecar that existed but would not parse made a PARTIAL track read as
complete, permanently — the viewer loses every cue past the cached window.
- An unreadable crop/trim/cut-list in an edit spec was dropped, so the clip came
out uncropped, untrimmed, or shorter than the one the user built.
- A dropped `max_height` downloaded at best quality against an explicit cap.
- A failed videos.list enrich made every candidate undroppable-from-judgement and
therefore dropped: a working search rendered as "no results".
- An unusable password hash answered as "wrong password", locking the owner out
and telling them they mistyped; a rotated TOKEN_ENCRYPTION_KEY turned every
stored token into "not connected".
- A stored admin setting that would not coerce fell back to the env default while
the settings page went on showing the saved value.
- A one-off InnerTube config failure pinned the whole process to the hard-coded
fallback key, and searches return nothing when that goes stale.
- The HLS sweep reported 0 swept — like a clean run — when it could not list its
own root, leaving every crashed session's segments on disk forever.
Four of them are counted rather than logged per item (dropped cues, dropped cut
list segments, unreclaimed GC files) and reported once per call, because the
alternative is one line per cue over a whole subtitle track.
ruff gains S110 and BLE001. Every blind `except Exception` that remains carries
`# noqa: BLE001 — <why>`, which is the point: catching everything is sometimes
right, and always worth a sentence. plex/sync.py's three best-effort backfill
catches now keep the traceback — a blind catch over a batch that fetches AND
writes could be Plex, the database or our own bug, and `%s` of the exception told
them apart badly. main.py's boot-time log-level read likewise.
logs.event's `code` is positional-only now: `event("x", code=...)` was a TypeError,
and `code` is exactly the word a caller reaches for when the thing that failed has
one of its own.
tests/test_swallowed_exceptions.py is the wider rule ruff cannot see: a handler
that neither logs nor re-raises must say why. Both cases mutation-checked — a
marker removed and a log line removed each fail it, naming the handler.
556 lines
25 KiB
Python
556 lines
25 KiB
Python
"""Download worker process entry point.
|
|
|
|
Runs in a DEDICATED container (`command: python -m app.worker`, `WORKER_ENABLED=1`), separate
|
|
from the API so long yt-dlp downloads + ffmpeg postprocessing never block web requests.
|
|
|
|
Design:
|
|
* N loop threads (download_worker_concurrency) each claim a `queued` DownloadJob with
|
|
FOR UPDATE SKIP LOCKED and process it.
|
|
* The shared MediaAsset is the dedup unit. A job whose asset is already `ready` finishes
|
|
instantly (no re-download); a fresh asset is claimed pending→downloading by exactly one
|
|
worker (a DB compare-and-set), so two jobs for the same video+format never double-download.
|
|
* Live progress is written to the job row (short-lived sessions, throttled ~1/s) — the
|
|
frontend polls it; the worker can't reach the API's in-process WebSocket registry.
|
|
* Pause/cancel are cooperative: the progress hook re-reads the job status and aborts the
|
|
yt-dlp download if it's no longer `running`.
|
|
* Crash-safe: on startup, orphaned `running` jobs are requeued and `downloading` assets reset.
|
|
"""
|
|
|
|
import os
|
|
import shutil
|
|
import signal
|
|
import subprocess
|
|
import sys
|
|
import threading
|
|
import time
|
|
from datetime import datetime, timezone
|
|
|
|
from sqlalchemy import text
|
|
|
|
from app import logs
|
|
from app import logsink
|
|
from app import sysconfig
|
|
from app.config import settings
|
|
from app.db import SessionLocal
|
|
from app.downloads import edit as editmod
|
|
from app.downloads import credentials, formats, quota, service
|
|
from app.downloads import errors as dlerrors
|
|
from app.downloads.jobdb import (
|
|
_edit_staging_dir,
|
|
_requeue_if_running,
|
|
_reset_asset_pending,
|
|
_staging_dir,
|
|
)
|
|
from app.downloads.status import STOPPED_BY_ROUTE, JobStatus
|
|
from app.models import DownloadJob, MediaAsset
|
|
|
|
log = logs.get_logger(__name__)
|
|
_stop = threading.Event()
|
|
|
|
|
|
def _handle_signal(signum, frame):
|
|
_stop.set()
|
|
|
|
|
|
# --- claim + recovery -----------------------------------------------------------------------
|
|
|
|
|
|
def _wait_for_schema(timeout: float = 300.0) -> bool:
|
|
"""Block until the download tables exist before doing any DB work. Returns False if they never
|
|
appeared (or we were asked to stop while waiting).
|
|
|
|
The API applies migrations on its OWN startup, and the worker is a separate container that may
|
|
boot first (they start in parallel). Without this, the worker would hit a missing `download_jobs`
|
|
table and crash-loop until the API's migrations land. Poll patiently instead — and if the schema
|
|
is still absent when the timeout runs out, say so and let the caller exit, because "continuing"
|
|
only moved the same crash one call further down (into `_recover_orphans`), where it read as an
|
|
unrelated failure."""
|
|
deadline = time.monotonic() + timeout
|
|
while not _stop.is_set():
|
|
try:
|
|
with SessionLocal() as db:
|
|
db.execute(text("SELECT 1 FROM download_jobs LIMIT 1"))
|
|
return True
|
|
except Exception: # noqa: BLE001 — table not migrated yet / DB not up yet
|
|
if time.monotonic() > deadline:
|
|
# The traceback, not `str(exc)[:120]`: this is the last thing the worker says
|
|
# before it exits, and which driver call failed is the whole question.
|
|
log.exception(
|
|
"download schema still absent after %ss — the API's migrations have not "
|
|
"landed; exiting so the container restarts",
|
|
timeout,
|
|
)
|
|
return False
|
|
log.info("waiting for the database schema (API migrations)…")
|
|
_stop.wait(3.0)
|
|
return False
|
|
|
|
|
|
def _recover_orphans() -> None:
|
|
"""Reset state left behind by a previous crash so nothing is stuck."""
|
|
with SessionLocal() as db:
|
|
db.execute(text("UPDATE download_jobs SET status='queued' WHERE status='running'"))
|
|
db.execute(text("UPDATE media_assets SET status='pending' WHERE status='downloading'"))
|
|
db.commit()
|
|
|
|
|
|
def _claim_job() -> int | None:
|
|
"""Claim the oldest queued job whose user is under their per-user concurrency limit AND whose
|
|
shared asset isn't already being downloaded by another worker.
|
|
|
|
The asset exclusion (`NOT EXISTS … media_assets.status='downloading'`) stops the same-asset ping-
|
|
pong (B-sub-3): without it, a second worker claims a queued job that shares an in-flight asset,
|
|
finds it busy, requeues it, and immediately re-claims it — burning a worker slot behind one
|
|
in-flight asset and starving everyone else. Excluded here, that job simply waits until its asset
|
|
settles, so `_process_job` no longer needs to sleep on the busy/lost-race paths.
|
|
|
|
Users already AT their concurrency limit are excluded IN THE SQL, so one user flooding the
|
|
queue can't monopolize the fetch window and starve everyone behind them. (The old fixed
|
|
LIMIT-50 window did exactly that: 50 queued jobs from a single capped user filled the window,
|
|
`chosen` came back None, and every other user's work — sitting past position 50 — was never
|
|
seen.) The per-user limit is the quota row's `max_concurrent` (else the sysconfig default);
|
|
`unlimited` users are never capped. A final Python `at_concurrency_limit` re-check still guards
|
|
the tiny race between the exclusion subquery and the claim (it can only ever reject, never
|
|
over-claim), so correctness never depends on the two observations agreeing."""
|
|
with SessionLocal() as db:
|
|
default_mc = sysconfig.get_int(db, "download_default_max_concurrent")
|
|
rows = db.execute(
|
|
text(
|
|
"""
|
|
SELECT id, user_id FROM download_jobs
|
|
WHERE status='queued'
|
|
AND NOT EXISTS (
|
|
SELECT 1 FROM media_assets a
|
|
WHERE a.id = download_jobs.asset_id AND a.status = 'downloading'
|
|
)
|
|
AND user_id NOT IN (
|
|
SELECT r.user_id
|
|
FROM download_jobs r
|
|
LEFT JOIN download_quotas q ON q.user_id = r.user_id
|
|
WHERE r.status='running' AND r.user_id IS NOT NULL
|
|
GROUP BY r.user_id, q.max_concurrent, q.unlimited
|
|
HAVING COALESCE(q.unlimited, false) = false
|
|
AND count(*) >= GREATEST(COALESCE(q.max_concurrent, :dmc), 1)
|
|
)
|
|
ORDER BY queue_pos, id
|
|
FOR UPDATE SKIP LOCKED
|
|
LIMIT 50
|
|
"""
|
|
),
|
|
{"dmc": default_mc},
|
|
).fetchall()
|
|
chosen = next((jid for jid, uid in rows if not quota.at_concurrency_limit(db, uid)), None)
|
|
if chosen is None:
|
|
db.commit() # release the row locks; nothing eligible right now
|
|
return None
|
|
db.execute(
|
|
text(
|
|
"UPDATE download_jobs SET status='running', phase='starting', updated_at=now() "
|
|
"WHERE id = :id"
|
|
),
|
|
{"id": chosen},
|
|
)
|
|
db.commit()
|
|
return chosen
|
|
|
|
|
|
def _claim_asset(asset_id: int) -> bool:
|
|
"""Compare-and-set pending→downloading. True if THIS worker won the right to download."""
|
|
with SessionLocal() as db:
|
|
row = db.execute(
|
|
text(
|
|
"UPDATE media_assets SET status='downloading' "
|
|
"WHERE id = :id AND status='pending' RETURNING id"
|
|
),
|
|
{"id": asset_id},
|
|
).fetchone()
|
|
db.commit()
|
|
return row is not None
|
|
|
|
|
|
# --- supervised child execution + watchdog --------------------------------------------------
|
|
# The heavy work runs in a child process so a wedged yt-dlp/ffmpeg is HARD-killable. A worker thread
|
|
# can't be killed, a C-level spin (e.g. a catastrophic regex) can't even be cooperatively cancelled,
|
|
# and one such thread starves the rest via the GIL — the incident that motivated this. The parent
|
|
# thread supervises: it kills the child (and its process group, so ffmpeg grandchildren die too) on
|
|
# a route pause/cancel, on a no-progress stall, or past an absolute wall-clock cap.
|
|
#
|
|
# POSIX-only: the kill path uses process groups (start_new_session / os.killpg / os.getpgid), so the
|
|
# worker is Linux-container-only — never run `python -m app.worker` natively on Windows.
|
|
# Edits ALSO carry a second, tighter watchdog inside run_ffmpeg (a per-ffmpeg stall timeout with a
|
|
# cleaner "ffmpeg stalled" error); this process-level supervisor is the uniform backstop for both the
|
|
# download and edit paths, and the last line of defence for anything run_ffmpeg can't see.
|
|
|
|
_CHILD_OK = 0
|
|
|
|
|
|
def _spawn_download_child(job_id: int) -> subprocess.Popen:
|
|
"""Start the child that does the actual download/edit for `job_id`, in its OWN session/process
|
|
group so stopping it also stops any ffmpeg / helper processes yt-dlp spawned."""
|
|
return subprocess.Popen(
|
|
[sys.executable, "-m", "app.downloads.runner", str(job_id)],
|
|
start_new_session=True,
|
|
)
|
|
|
|
|
|
def _job_liveness(job_id: int) -> tuple[str | None, float | None]:
|
|
"""(status, seconds-since-last-row-update) for the job, computed with the DB clock so there is
|
|
no parent↔DB skew. A live download rewrites its row ~1/s; a stalled one never does. Returns
|
|
(None, None) if the row is gone."""
|
|
with SessionLocal() as db:
|
|
row = db.execute(
|
|
text(
|
|
"SELECT status, EXTRACT(EPOCH FROM (now() - updated_at)) "
|
|
"FROM download_jobs WHERE id = :id"
|
|
),
|
|
{"id": job_id},
|
|
).one_or_none()
|
|
if row is None:
|
|
return None, None
|
|
return row[0], (float(row[1]) if row[1] is not None else None)
|
|
|
|
|
|
def _signal_group(proc: subprocess.Popen, pgid: int | None, sig: int) -> None:
|
|
"""Signal the child's whole process group (falling back to the child alone), swallowing the race
|
|
where it has already exited."""
|
|
try:
|
|
if pgid is not None:
|
|
os.killpg(pgid, sig)
|
|
else:
|
|
proc.send_signal(sig)
|
|
except ProcessLookupError:
|
|
pass # silent-ok: the child already exited — the race this function is written around
|
|
except OSError as exc:
|
|
# Any OTHER refusal means the signal did not land and the child is still running, which
|
|
# is how a "cancelled" download keeps writing to a staging tree nobody is watching.
|
|
log.info(
|
|
"a download child could not be signalled",
|
|
extra=logs.event(
|
|
"download.child.signal_failed", signal=int(sig), reason=type(exc).__name__
|
|
),
|
|
)
|
|
|
|
|
|
def _stop_child(proc: subprocess.Popen, grace_s: float) -> None:
|
|
"""SIGTERM the child's process GROUP, then SIGKILL if it outlives the grace period. Shares the
|
|
terminate→wait→kill escalation with run_ffmpeg (`editmod.escalate_kill`); the only difference is
|
|
that we signal the whole group so ffmpeg grandchildren die too. `escalate_kill` no-ops if the
|
|
child already exited, and `getpgid` yields None for a reaped pid, so no separate poll guard."""
|
|
try:
|
|
pgid: int | None = os.getpgid(proc.pid)
|
|
# silent-ok: `getpgid` refuses for a reaped pid, which the docstring above already treats as
|
|
# the ordinary case — the escalation then signals the child alone and no-ops if it is gone.
|
|
except OSError:
|
|
pgid = None
|
|
editmod.escalate_kill(
|
|
proc,
|
|
terminate=lambda: _signal_group(proc, pgid, signal.SIGTERM),
|
|
kill=lambda: _signal_group(proc, pgid, signal.SIGKILL),
|
|
grace_s=grace_s,
|
|
label="download child",
|
|
)
|
|
|
|
|
|
def _stop_and_free(job_id: int, asset_id: int, proc: subprocess.Popen, grace_s: float) -> None:
|
|
"""Kill the child, free its claimed asset, and clear its staging — the common tail of every path
|
|
that stops a child (a route pause/cancel, a shutdown, an over-cap abandon)."""
|
|
_stop_child(proc, grace_s)
|
|
_reset_asset_pending(asset_id)
|
|
_cleanup_staging(asset_id)
|
|
|
|
|
|
def _fail_after_kill(
|
|
job_id: int, asset_id: int, proc: subprocess.Popen, grace_s: float, error: str, error_code: str
|
|
) -> None:
|
|
"""`_stop_and_free` + mark the job errored — for the watchdog's stall/timeout kills and the
|
|
supervisor-error fallback."""
|
|
_stop_and_free(job_id, asset_id, proc, grace_s)
|
|
_fail_running(job_id, error=error, error_code=error_code)
|
|
|
|
|
|
def _fail_running(
|
|
job_id: int, *, error: str, error_code: str | None, error_detail: str | None = None
|
|
) -> None:
|
|
"""Fail a claimed job — only while it is still `running` (`service.finalize_if_running`), so a
|
|
cancel/pause a route landed first is never turned back into `error`, which holds the asset the
|
|
route already released. `error_detail` is written every time: a failure never shows an earlier
|
|
failure's explanation."""
|
|
with SessionLocal() as db:
|
|
service.finalize_if_running(
|
|
db,
|
|
job_id,
|
|
status=JobStatus.ERROR,
|
|
error=error,
|
|
error_code=error_code,
|
|
error_detail=error_detail,
|
|
phase=None,
|
|
)
|
|
db.commit()
|
|
|
|
|
|
def _run_supervised(job_id: int, asset_id: int) -> None:
|
|
"""Run the heavy download/edit for an already-claimed asset in a child process and watchdog it.
|
|
|
|
A clean child exit means the child already wrote the terminal job/asset state (done/error, or a
|
|
cooperative pause via the in-child progress hook) — we leave it. We step in only when the child
|
|
can't finish or self-abort: a route paused/cancelled it (kill, free the asset, keep the route's
|
|
status), it made no progress for `stall_timeout` or ran past `wall_clock_max` (kill, fail the
|
|
job), the worker is shutting down (kill, requeue), or it crashed without setting state."""
|
|
poll_s = max(1, settings.download_supervisor_poll_seconds)
|
|
stall_s = settings.download_stall_timeout_seconds
|
|
wall_s = settings.download_wall_clock_max_seconds
|
|
grace_s = settings.download_kill_grace_seconds
|
|
|
|
try:
|
|
proc = _spawn_download_child(job_id)
|
|
except Exception: # noqa: BLE001 — spawn failed; don't leave the asset stuck 'downloading'
|
|
log.exception("failed to spawn download child for job %s", job_id)
|
|
_reset_asset_pending(asset_id)
|
|
_requeue_if_running(job_id, None)
|
|
return
|
|
|
|
started = time.monotonic()
|
|
reason: str | None = None
|
|
try:
|
|
while True:
|
|
try:
|
|
proc.wait(timeout=poll_s)
|
|
break # child finished on its own
|
|
# silent-ok: the timeout is the loop's clock, not a failure — it is what gives this
|
|
# thread a turn to check whether a route cancelled the job.
|
|
except subprocess.TimeoutExpired:
|
|
pass
|
|
status, stale = _job_liveness(job_id)
|
|
if status is None or status in STOPPED_BY_ROUTE:
|
|
reason = "cancelled" # a route stopped it (or the row vanished)
|
|
break
|
|
if status == "running":
|
|
if stale is not None and stale > stall_s:
|
|
reason = "stalled"
|
|
break
|
|
if _stop.is_set():
|
|
reason = "shutdown"
|
|
break
|
|
# Absolute wall-clock cap frees the worker slot even for a FINISHED child wedged in
|
|
# teardown. A still-running job over the cap is failed ("timeout"); a terminal one (already
|
|
# done/error) just gets its lingering child killed, its DB status left intact ("abandon").
|
|
if time.monotonic() - started > wall_s:
|
|
reason = "timeout" if status == "running" else "abandon"
|
|
break
|
|
# else: a terminal status under the cap (done/error, or a re-queued contended asset) means
|
|
# the child is exiting on its own — keep polling; NEVER kill a finishing child (that could
|
|
# drop its completion notice).
|
|
except Exception: # noqa: BLE001 — a watchdog error must NEVER leave the child running unsupervised
|
|
log.exception("supervisor loop failed for job %s; stopping the child", job_id)
|
|
_fail_after_kill(
|
|
job_id,
|
|
asset_id,
|
|
proc,
|
|
grace_s,
|
|
"The download was stopped after a worker error.",
|
|
"unknown",
|
|
)
|
|
return
|
|
|
|
if reason is None:
|
|
# Child exited. Reconcile the asset we claimed: free it if the child left it non-finalized —
|
|
# a crash/OOM, or an early return on a job deleted mid-flight — a no-op when it finalized
|
|
# normally. Then, if it died non-zero while the job is still 'running', fail the job so it is
|
|
# not left stuck. Guarded: the child is already dead, so a DB blip here must not propagate.
|
|
try:
|
|
_reset_asset_pending(asset_id)
|
|
if proc.returncode != _CHILD_OK:
|
|
status, _ = _job_liveness(job_id)
|
|
if status == "running":
|
|
_cleanup_staging(asset_id)
|
|
_fail_running(
|
|
job_id,
|
|
error=f"The download process exited unexpectedly (code {proc.returncode}).",
|
|
error_code="unknown",
|
|
)
|
|
log.warning(
|
|
"job %s: child exited %s without finishing", job_id, proc.returncode
|
|
)
|
|
except Exception: # noqa: BLE001 — boot recovery is the backstop for a stuck 'running' job
|
|
log.exception("post-exit reconciliation failed for job %s", job_id)
|
|
return
|
|
|
|
if reason == "stalled":
|
|
_fail_after_kill(
|
|
job_id,
|
|
asset_id,
|
|
proc,
|
|
grace_s,
|
|
f"The download stalled (no progress for {int(stall_s)}s) and was stopped.",
|
|
"network",
|
|
)
|
|
return
|
|
if reason == "timeout":
|
|
_fail_after_kill(
|
|
job_id,
|
|
asset_id,
|
|
proc,
|
|
grace_s,
|
|
f"The download exceeded the {int(wall_s)}s time limit and was stopped.",
|
|
"network",
|
|
)
|
|
return
|
|
# cancelled / abandon / shutdown: stop + free the asset, but leave the job's own DB status.
|
|
_stop_and_free(job_id, asset_id, proc, grace_s)
|
|
if reason == "shutdown":
|
|
_requeue_if_running(job_id, None) # resume on the next boot
|
|
elif reason == "abandon":
|
|
log.warning(
|
|
"job %s: killed a child wedged past the wall-clock; DB status left intact", job_id
|
|
)
|
|
else: # cancelled
|
|
log.info("job %s: child stopped (route pause/cancel)", job_id)
|
|
|
|
|
|
def _cleanup_staging(asset_id: int) -> None:
|
|
"""Best-effort removal of a killed child's leftover staging dirs (its own `finally` never ran).
|
|
The next attempt on this asset also wipes staging on start, so this is just tidiness/disk."""
|
|
for path in (_staging_dir(asset_id), _edit_staging_dir(asset_id)):
|
|
shutil.rmtree(path, ignore_errors=True)
|
|
|
|
|
|
# --- processing -----------------------------------------------------------------------------
|
|
|
|
|
|
def _process_job(job_id: int) -> None:
|
|
"""Dispatch a claimed job. The trivial outcomes (cache hit, dead/contended asset) are handled
|
|
inline and cheaply; the actual heavy work (yt-dlp download / ffmpeg edit) is claimed here and
|
|
then run in a supervised CHILD PROCESS (`_run_supervised`) so a wedged yt-dlp/ffmpeg can be
|
|
hard-killed — a worker THREAD cannot be killed, nor (when it spins in a C call) even
|
|
cooperatively cancelled, and one such thread starves the others via the GIL."""
|
|
with SessionLocal() as db:
|
|
job = db.get(DownloadJob, job_id)
|
|
if job is None:
|
|
return
|
|
asset = db.get(MediaAsset, job.asset_id) if job.asset_id else None
|
|
asset_id = asset.id if asset else None
|
|
asset_status = asset.status if asset else None
|
|
asset_error = asset.error if asset else None
|
|
if asset_status == "error":
|
|
# A job that meets an already-failed asset gets that failure's explanation — not a bare
|
|
# "Download failed". Only the text is on the asset, so it is diagnosed again, for this
|
|
# link, with whether a cookie goes with it asked of the same gate the download child
|
|
# used (`credentials.usable_cookie`) — as it stands now, so the advice fits this job's Retry.
|
|
url = service.source_url(job.source_kind, job.source_ref)
|
|
cookie = credentials.usable_cookie(
|
|
db, job.user_id, url, asset, formats.normalize(job.profile_snapshot)
|
|
)
|
|
diagnosis = dlerrors.diagnose(asset_error, url=url, cookie_attached=cookie is not None)
|
|
del cookie # only whether there is one matters here
|
|
|
|
if asset is None:
|
|
_fail_running(job_id, error="No media asset", error_code=None)
|
|
return
|
|
|
|
if asset_status == "ready":
|
|
_finish_from_cache(job_id, asset_id)
|
|
return
|
|
if asset_status == "error":
|
|
_fail_running(
|
|
job_id,
|
|
error=asset_error or "Download failed",
|
|
error_code=diagnosis.code,
|
|
error_detail=diagnosis.detail,
|
|
)
|
|
return
|
|
if asset_status == "downloading":
|
|
# Another worker won this asset between our claim and here (a narrow race — the claim SQL
|
|
# excludes downloading-asset jobs). Requeue and let it be reclaimed once the asset settles;
|
|
# no sleep — the claim exclusion keeps it out of the window until then, so no busy-loop.
|
|
_requeue_if_running(job_id, "waiting")
|
|
return
|
|
|
|
# asset_status == "pending": try to win the download / edit.
|
|
if not _claim_asset(asset_id):
|
|
# Lost the pending->downloading race; requeue (same reasoning — the exclusion prevents a spin).
|
|
_requeue_if_running(job_id, "waiting")
|
|
return
|
|
|
|
# The asset is now ours ('downloading'): run the heavy work in a supervised child process.
|
|
_run_supervised(job_id, asset_id)
|
|
|
|
|
|
def _finish_from_cache(job_id: int, asset_id: int) -> None:
|
|
with SessionLocal() as db:
|
|
job = db.get(DownloadJob, job_id)
|
|
asset = db.get(MediaAsset, asset_id)
|
|
if job and asset:
|
|
fields = {"status": JobStatus.DONE, "progress": 100, "phase": None}
|
|
if not job.display_name and asset.title:
|
|
# display_name is VARCHAR(255); a title can be longer (e.g. a recipe in the title).
|
|
fields["display_name"] = asset.title[:255]
|
|
asset.last_access_at = datetime.now(timezone.utc)
|
|
# Atomic finalize: only marks the job done while it is still running, so a concurrent
|
|
# cancel/pause is never resurrected (see service.finalize_if_running).
|
|
service.finalize_if_running(db, job_id, **fields)
|
|
db.commit()
|
|
|
|
|
|
# --- loop -----------------------------------------------------------------------------------
|
|
|
|
|
|
def _worker_loop(idx: int) -> None:
|
|
while not _stop.is_set():
|
|
try:
|
|
job_id = _claim_job()
|
|
except Exception: # noqa: BLE001
|
|
log.exception("claim failed")
|
|
_stop.wait(settings.download_poll_seconds)
|
|
continue
|
|
if job_id is None:
|
|
_stop.wait(settings.download_poll_seconds)
|
|
continue
|
|
# Tag every line this job produces (including the crash below) with the job id: a worker runs
|
|
# N jobs at once on N threads, so without it the interleaved output is unreadable. The bind
|
|
# wraps the handler too — bound inside the `try`, it was already reset by the time the crash
|
|
# line was logged.
|
|
with logs.bind(job_id=job_id):
|
|
try:
|
|
_process_job(job_id)
|
|
except Exception: # noqa: BLE001
|
|
log.exception("processing job %s crashed", job_id)
|
|
_fail_running(job_id, error="Worker error", error_code=None)
|
|
|
|
|
|
def main() -> None:
|
|
logs.configure_logging("worker")
|
|
signal.signal(signal.SIGTERM, _handle_signal)
|
|
signal.signal(signal.SIGINT, _handle_signal)
|
|
|
|
if not settings.worker_enabled:
|
|
log.info("WORKER_ENABLED is off; worker exiting (nothing to do).")
|
|
return
|
|
|
|
# The API may still be applying migrations when we boot. If they never arrive, exit instead of
|
|
# pressing on into a guaranteed crash — with a FAILURE status, so the container comes back under
|
|
# any restart policy (`on-failure` too, not just the `unless-stopped` our composes ship) and the
|
|
# exit code doesn't claim success for a worker that never started working.
|
|
if not _wait_for_schema():
|
|
if _stop.is_set():
|
|
return # asked to shut down while waiting — that IS a clean exit
|
|
raise SystemExit(1)
|
|
# After the wait, not before: the DB is reachable now. The wait checks `download_jobs`, not
|
|
# `log_events`, so on the deploy that first adds the table the worker may start before the API
|
|
# has migrated — those few seconds of batches fail and say so on stderr; nothing else breaks.
|
|
logsink.start()
|
|
_recover_orphans()
|
|
n = max(1, settings.download_worker_concurrency)
|
|
log.info("Download worker started (concurrency=%d, root=%s).", n, settings.download_root)
|
|
threads = [threading.Thread(target=_worker_loop, args=(i,), daemon=True) for i in range(n)]
|
|
for t in threads:
|
|
t.start()
|
|
while not _stop.is_set():
|
|
_stop.wait(1.0)
|
|
log.info("Download worker stopping…")
|
|
for t in threads:
|
|
t.join(timeout=5)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
main()
|