feat(telemetry): readable Sentry errors, Python tracebacks, and diagnostic mode

Keeps a real, redacted error message instead of "Error: Error", surfaces Python tracebacks in Sentry as a vetted context, and adds an opt-in SNAPOTTER_SENTRY_DIAGNOSTIC verbose mode plus SNAPOTTER_SENTRY_DSN_OVERRIDE. The default fleet path ships nothing on the never-collect list; raw detail is reachable only via the opt-in flag. Also classifies Redis OOM/READONLY replies as operational and removes a ReDoS in stack-frame extraction.
This commit is contained in:
SnapOtter
2026-08-03 13:13:38 +08:00
committed by GitHub
parent 865390ce66
commit 04ef1141fb
22 changed files with 593 additions and 178 deletions
+9 -3
View File
@@ -27,6 +27,8 @@ import io
import os
import traceback
from sidecar_errors import build_error_envelope
# ── Optional OpenTelemetry tracing (enterprise only) ─────────────
_tracer = None
@@ -294,10 +296,14 @@ def _run_script_main(script_name, args):
except SystemExit as e:
exit_code = e.code if isinstance(e.code, int) else 1
except Exception as e:
# Log full traceback to stderr for diagnostics
# Log full traceback to stderr for local diagnostics (unchanged).
traceback.print_exc(file=sys.stderr)
# Write error to the captured stdout
sys.stdout.write(json.dumps({"success": False, "error": str(e)}) + "\n")
info = build_error_envelope(e)
# Keep `error` as the redacted string for back-compatible consumers;
# add `errorInfo` (type + our frames) for the structured Sentry path.
sys.stdout.write(
json.dumps({"success": False, "error": info["message"], "errorInfo": info}) + "\n"
)
sys.stdout.flush()
exit_code = 1
finally:
+78
View File
@@ -0,0 +1,78 @@
"""
Structured error envelope for the sidecar. Mirrors the Node redactMessage so a
Python failure reaches Sentry with its type, a redacted message, and our own
stack frames (basename only) instead of a bare "Error: Error".
"""
import os
import re
import traceback
# _CTRL: matches ASCII control chars 0x00-0x1F and 0x7F only. Written with \x
# hex escapes so no literal control bytes appear in this file. It must NOT match
# spaces, "/", ".", or any printable character.
_CTRL = re.compile(r"[\x00-\x1f\x7f]")
_BLOB = re.compile(r"blob:[^\s\"')]+")
_DATA = re.compile(r"data:[^\s\"')]+")
_URL = re.compile(r"https?://[^\s\"')]+")
_PATH = re.compile(r"(?:/(?:Users|home|root|data|tmp|var|app|opt|mnt|srv)|[A-Za-z]:\\)[^\s\"')]*")
# Relative object-storage keys (uploads/<jobId>/…, outputs/…, previews/…) carry a
# user-supplied filename tail; mask them like absolute paths. Runs after _PATH,
# which already swallows the absolute /data/uploads/… form.
_RELKEY = re.compile(r"\b(?:uploads|outputs|previews)/[^\s\"')]+")
_IP = re.compile(r"\b\d{1,3}(?:\.\d{1,3}){3}\b")
# IPv6: a full 8-group form, or any ::-compressed form (::1, fe80::…, …::). The
# negative lookbehind/lookahead ((?<![\w:]) … (?![\w:])) require the address to
# stand alone, so C++/Rust scope resolution (std::bad_alloc, core::result) is left
# intact. A plain decimal version like 2.2.0 has no colons, and a bare HH:MM needs
# no ::, so both survive too.
_IPV6 = re.compile(
r"(?<![\w:])(?:(?:[0-9a-fA-F]{1,4}:){7}[0-9a-fA-F]{1,4}|"
r"(?:[0-9a-fA-F]{1,4}(?::[0-9a-fA-F]{1,4})*)?::(?:[0-9a-fA-F]{1,4}(?::[0-9a-fA-F]{1,4})*)?)(?![\w:])"
)
_EMAIL = re.compile(r"\b[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Za-z]{2,}\b")
_USER_FILE_EXT = (
"jpe?g|png|gif|webp|avif|heif?|tiff?|bmp|svg|raw|psd|mp4|mov|avi|mkv|webm|"
"flv|wmv|m4v|mp3|wav|flac|aac|ogg|m4a|opus|pdf|docx?|xlsx?|pptx?|odt|ods|"
"odp|txt|csv|epub|zip"
)
_FILE = re.compile(r"\b[\w-]{1,80}\.(?:" + _USER_FILE_EXT + r")\b", re.IGNORECASE)
_QUOTED = re.compile(r"(['\"])(.{24,}?)\1")
_HEX = re.compile(r"\b[0-9a-fA-F]{16,}\b")
_MAX_LEN = 300
_SIDECAR_DIR = os.path.dirname(os.path.abspath(__file__))
def redact(message):
s = _CTRL.sub(" ", str(message or ""))
s = _BLOB.sub("<blob>", s)
s = _DATA.sub("<data>", s)
s = _URL.sub("<url>", s)
s = _PATH.sub("<path>", s)
s = _RELKEY.sub("<path>", s)
s = _IP.sub("<ip>", s)
s = _IPV6.sub("<ip>", s)
s = _EMAIL.sub("<email>", s)
s = _QUOTED.sub(lambda m: m.group(1) + "<value>" + m.group(1), s)
s = _HEX.sub("<hex>", s)
s = _FILE.sub("<file>", s)
s = re.sub(r"\s+", " ", s).strip()
return (s[:_MAX_LEN] + "") if len(s) > _MAX_LEN else s
def _our_frames(exc):
frames = []
for fr in traceback.extract_tb(exc.__traceback__):
# Keep only our sidecar-script frames; drop stdlib and venv/site-packages.
if os.path.dirname(os.path.abspath(fr.filename)) != _SIDECAR_DIR:
continue
frames.append({"file": os.path.basename(fr.filename), "line": fr.lineno, "func": fr.name})
return frames[-20:]
def build_error_envelope(exc):
"""A JSON-serializable {type, message, frames} describing a caught exception."""
return {
"type": type(exc).__name__,
"message": redact(str(exc)),
"frames": _our_frames(exc),
}
+31
View File
@@ -0,0 +1,31 @@
import os
import sys
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
from sidecar_errors import build_error_envelope, redact # noqa: E402
def test_redact_masks_paths_and_files():
assert redact("open /data/uploads/9f/in.bin") == "open <path>"
assert redact("cannot read family_photo.JPG") == "cannot read <file>"
assert redact("torch 2.2.0 ok") == "torch 2.2.0 ok"
def test_envelope_shape_and_frames():
try:
raise RuntimeError("CUDA out of memory for /data/x.png")
except RuntimeError as exc:
env = build_error_envelope(exc)
assert env["type"] == "RuntimeError"
assert env["message"] == "CUDA out of memory for <path>"
assert isinstance(env["frames"], list) and len(env["frames"]) >= 1
top = env["frames"][-1]
assert top["file"] == "test_sidecar_errors.py"
assert isinstance(top["line"], int)
assert top["func"] == "test_envelope_shape_and_frames"
if __name__ == "__main__":
test_redact_masks_paths_and_files()
test_envelope_shape_and_frames()
print("ok")