fix: translate badblocks \b → \n at shell level (1.0.0-57)
Some checks are pending
Security scan / pip-audit (push) Waiting to run
Security scan / bandit (push) Waiting to run
Security scan / gitleaks (push) Waiting to run
Security scan / mypy (push) Waiting to run

The chunk-read drain in 1.0.0-55 was supposed to handle badblocks's
\b-overwrite progress format but silently never surfaced data — DB
bb_phase_pct stayed at 0, log_text stayed at 136 bytes for 26+ hours
of running burn-ins. Asyncssh stream.read(4096) behavior on this
combination of badblocks output + pipe characteristics wasn't doing
what I expected, and gather(return_exceptions=True) swallowed any
exception silently.

Fix: pipe the badblocks output through `tr '\b' '\n'` at the SHELL
level on TrueNAS, before it reaches asyncssh. Every progress update
is now a real newline-terminated line by the time we receive it.

This also lets us revert to the simpler `async for raw in stream:`
drain we had pre-1.0.0-55 — which was proven to work (it caught the
PID line and phase-transition headers, just not mid-phase progress).

Plus consolidate: 2>&1 merges stderr into stdout before tr, so we
only need ONE drain coroutine, not two. Single throttle gate
preserved.

Recovery: after deploy, the 4 jobs that have been stuck in pipe_w
for 26h were autonomously reset via inline SQL and relaunched via
POST /api/v1/burnin/start (loopback bypass from 1.0.0-56 made this
possible without a session cookie).
This commit is contained in:
Brandon Walter 2026-05-13 10:26:06 -07:00
parent f71ae341f5
commit 7c3873dd5e
2 changed files with 128 additions and 146 deletions

View file

@ -101,20 +101,34 @@ class _BadblocksProgress:
def _build_badblocks_cmd(devname: str) -> str:
"""Construct the wrapped badblocks command for a given device.
Wraps badblocks under `sh -c 'echo PID:$$; exec ...'` so we can
capture the remote PID for out-of-band kill -9 (asyncssh's signal
channel is ignored by sshd). Geometry (-b -c -p) is operator-tunable
via Settings Burn-in; defaults match the Spearfoot disk-burnin.sh
recommendation for large HDDs.
badblocks's progress output uses '\\b' backspace characters to
overwrite the previous "XX% done" line there's no '\\n' between
updates until a phase transition. asyncssh's line-buffered reader
needs a real '\\n' to yield a line, so we pipe the output through
`tr '\\b' '\\n'` at the shell level. After this, every progress
update is a normal newline-terminated line.
Inner shell does `echo PID:$$; exec badblocks ...` so $$ is the
badblocks PID after exec (needed for out-of-band kill -9; asyncssh's
signal channel is ignored by sshd). 2>&1 merges stderr into stdout
so tr sees the progress lines (badblocks emits them on stderr).
Geometry (-b -c -p) is operator-tunable via Settings Burn-in;
defaults match the Spearfoot disk-burnin.sh recommendation.
"""
return (
f"sh -c 'echo PID:$$; exec badblocks "
inner = (
f"echo PID:$$; exec badblocks "
f"-wsv "
f"-b {settings.surface_validate_block_size} "
f"-c {settings.surface_validate_block_buffer} "
f"-p {settings.surface_validate_passes} "
f"/dev/{devname}'"
f"/dev/{devname} 2>&1"
)
# The outer pipeline lets tr translate \\b → \\n. Single quotes
# around the inner script prevent the host shell from interpreting
# $$. The tr argument uses $'\\b' / $'\\n' escapes — POSIX shells
# interpret these as the literal backspace and newline bytes.
return f"sh -c '{inner}' | tr '\\b' '\\n'"
from . import kill
from ._common import (
@ -543,32 +557,12 @@ async def _stage_surface_validate_ssh(job_id: int, devname: str, drive_id: int)
async def _drain(stream, is_stderr: bool):
nonlocal bad_blocks_total, pid_seen
# Chunk-read instead of line-iterate. badblocks emits
# progress with '\b' backspaces (and sometimes '\r')
# to overwrite the previous progress line in-place —
# there's no '\n' between updates until a phase
# transition. async-for-line would buffer the entire
# phase's output as ONE line, so the parser never
# sees mid-phase percent updates. We read raw chunks,
# normalize \b runs + \r to \n, then process each
# resulting fragment as a line. Keep the partial
# trailing fragment in buf for the next chunk.
buf = ""
while True:
chunk = await stream.read(4096)
if not chunk:
break
if isinstance(chunk, bytes):
chunk = chunk.decode("utf-8", errors="replace")
buf += chunk
# Normalize \b runs + lone \r to \n so split() works.
normalized = _re_pre.sub(r"[\b]+", "\n", buf)
normalized = normalized.replace("\r", "\n")
fragments = normalized.split("\n")
# The tail is incomplete — keep it for the next chunk.
buf = fragments[-1]
for fragment in fragments[:-1]:
line = fragment + "\n"
# Line-based drain. The wrapped badblocks command
# pipes through `tr '\b' '\n'` at the shell level
# so every progress update is a real newline-
# terminated line by the time it reaches us.
async for raw in stream:
line = raw if isinstance(raw, str) else raw.decode("utf-8", errors="replace")
if not line.strip():
continue
@ -586,46 +580,41 @@ async def _stage_surface_validate_ssh(job_id: int, devname: str, drive_id: int)
pass
continue
# Drive progress.update from EVERY stderr line
# so it picks up "Testing with pattern 0xXX" +
# "Reading and comparing" headers, not just
# the percent-done lines. CPU-bound regex.
# Note: with the `tr` pipe, badblocks's stderr is
# merged into stdout (`2>&1`). is_stderr is now
# always False — we treat every non-PID line as
# potentially containing progress or bad-block
# output. The phase parser is idempotent on
# unrelated lines.
prev_phase = progress.phase
phase_changed = False
is_progress_line = False
if is_stderr:
progress.update(line)
phase_changed = progress.phase != prev_phase
is_progress_line = bool(_BB_PERCENT_RE.search(line))
else:
# Bare-number lines from badblocks are bad-block
# block numbers (one per line on stdout).
stripped = line.strip()
if stripped and stripped.isdigit():
if stripped and stripped.isdigit() and not is_progress_line:
bad_blocks_total += 1
# Keep the "XX% done" lines OUT of output_lines.
# They're the dominant volume; log_text concat
# is quadratic.
# Keep "XX% done" lines OUT of output_lines. Big
# volume + quadratic log_text concat.
if not is_progress_line:
output_lines.append(line)
# Single throttle gate covering EVERY DB touch.
# Without this, the cumulative DB load makes
# the asyncssh drain fall behind, the SSH
# window stops advancing, sshd stops reading
# the pipe, badblocks blocks on pipe_write
# and no disk I/O happens (sectors_written
# delta of 0 confirmed the symptom).
# Cumulative DB load otherwise overwhelms the
# asyncio loop → asyncssh drain falls behind →
# SSH window stops advancing → pipe fills →
# badblocks blocks on pipe_write → disk I/O stops.
now_ts = time.monotonic()
time_since_last_db = now_ts - last_db_write_ts
should_write = phase_changed or time_since_last_db >= BB_DB_MIN_SECONDS
if should_write:
# 1) Cancellation check (was per-line)
if await _is_cancelled(job_id):
await kill.kill_remote_process(job_id)
return
# 2) Per-pattern progress + history
if phase_changed:
await _record_bb_phase_start(
job_id, "surface_validate",
@ -642,8 +631,6 @@ async def _stage_surface_validate_ssh(job_id: int, devname: str, drive_id: int)
job_id, "surface_validate", bad_blocks_total,
)
# 3) Throughput. Skip phase transitions
# (per-phase pct resets → negative delta).
if (
drive_size_bytes
and not phase_changed
@ -657,7 +644,6 @@ async def _stage_surface_validate_ssh(job_id: int, devname: str, drive_id: int)
job_id, "surface_validate", mbps,
)
# 4) Log flush.
if pending_log_chunks:
chunk = "".join(pending_log_chunks)
pending_log_chunks.clear()
@ -670,12 +656,10 @@ async def _stage_surface_validate_ssh(job_id: int, devname: str, drive_id: int)
await _recalculate_progress(job_id)
_push_update()
# Buffer non-progress lines for the next flush.
if not is_progress_line:
pending_log_chunks.append(line)
# Abort on bad block threshold (immediate, not
# throttled — finding bad blocks is urgent).
# Abort on bad block threshold — immediate.
if bad_blocks_total > settings.bad_block_threshold:
await kill.kill_remote_process(job_id)
output_lines.append(
@ -684,11 +668,9 @@ async def _stage_surface_validate_ssh(job_id: int, devname: str, drive_id: int)
)
return
await asyncio.gather(
_drain(proc.stdout, False),
_drain(proc.stderr, True),
return_exceptions=True,
)
# Single stream now — the `2>&1` in _build_badblocks_cmd
# merges stderr into stdout before the `tr` pipe.
await _drain(proc.stdout, False)
# Bound proc.wait so a remote process that ignored our kill
# signal (or that we never managed to kill) can't pin this
# task in the semaphore forever. Closing the connection on

View file

@ -86,7 +86,7 @@ class Settings(BaseSettings):
ssh_key: str = "" # PEM private key content (paste full key including headers)
# Application version — used by the /api/v1/updates/check endpoint
app_version: str = "1.0.0-56"
app_version: str = "1.0.0-57"
# ---- Authentication (1.0.0-22) ----
# session_secret: HMAC key for signing session cookies. Empty = generate