fix(docker): enable PYTHONFAULTHANDLER for native-crash diagnostics

In-process native code -- pymupdf's classify/metadata `pymupdf.open` and
embedded Qdrant -- can segfault the interpreter during indexing. With only
PYTHONUNBUFFERED set, such a fault makes the container exit 139 (SIGSEGV) /
133 (SIGTRAP) with no logs at all, which is exactly what #926 reports on a
CPU-constrained self-hosted VPS.

Enabling PYTHONFAULTHANDLER makes the interpreter dump a Python + C-level
traceback to stderr on SIGSEGV/SIGABRT/SIGFPE/SIGBUS, so the faulting library
is identifiable from container logs. The handler is dormant during normal
operation (no output, negligible overhead).

Refs #926

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
Chris Coutinho
2026-06-18 17:59:40 +02:00
co-authored by Claude Opus 4.8
parent ccc3f42640
commit 81f8f9efe2
+7
View File
@@ -23,6 +23,13 @@ COPY . .
RUN uv sync --locked --no-dev --no-editable --no-cache --extra postgres
ENV PYTHONUNBUFFERED=1
# Dump a Python + C-level traceback to stderr on a fatal native fault
# (SIGSEGV/SIGABRT/SIGFPE/SIGBUS). In-process native code -- pymupdf's
# classify/metadata open and embedded Qdrant -- can segfault the interpreter
# during indexing, and without faulthandler the container just exits 139/133
# with no logs (see issue #926). The handler is cheap and writes nothing in
# normal operation.
ENV PYTHONFAULTHANDLER=1
ENV VIRTUAL_ENV=/app/.venv
ENV PATH=/app/.venv/bin:$PATH
ENV TESSDATA_PREFIX=/usr/share/tesseract-ocr/5/tessdata