docker-library / docker-library/python

Debbuging a `-slim` image?

Aperta
#903 1 commento 0 reazioni 0 assegnatari Vedi su GitHub

Nessuno ha ancora preso questa issue.

Lingua principale
Dockerfile
Stelle
2.8k
Fork
1.1k
Merge medio
3g 1h
PR unite (30g)
1

Descrizione

We are hitting a rare deadlock in production that can't be reproduced using debug images.
Only way to debug it is to attach to the production image as the problem happens.
Problem is that production are based on the -slim images.

apt install gdb and python3-dbg is not enough because the python running is built from source and not aligned with what is available in apt.

I've tried to start the corresponding non-slim image but their binaries don't seem to align enough to make gdb happy:

Start python inside -slim image

➡️ My end goal is to run gdb py-bt on this process:
docker run -ti python:3.11-slim-bookworm python3 -c "import time; time.sleep(1000)"

Inside this image there is no chance to debug at all due to missing gdb and debug symbols. This is expected.

Build a debugger image from the corresponding non-slim image.
FROM python:3.11-bookworm

RUN --mount=type=cache,target=/var/cache/apt --mount=type=cache,target=/var/lib/apt \
    apt update -y && \
    apt install -y gdb

docker build -t mydebugpython .

Use the newly built debug image to attach into the first running container

Find the container id and use that to attach to the same pid-namespace.

docker run \
    --cap-add SYS_PTRACE \
    --pid container:15aab7ea5f57 \
    --privileged  \
    -ti \
    --entrypoint bash docker.io/library/mydebugpython
root@bb90d04f8722:/# gdb -q python --pid 1 -ex "py-bt"

Reading symbols from python...
Attaching to program: /usr/local/bin/python, process 1

warning: Build ID mismatch between current exec-file /usr/local/bin/python
and automatically determined exec-file /usr/local/bin/python3.11
exec-file-mismatch handling is currently "ask"

Load new symbol table from "/usr/local/bin/python3.11"? (y or n) y
Reading symbols from /usr/local/bin/python3.11...
Reading symbols from target:/usr/local/bin/../lib/libpython3.11.so.1.0...
(No debugging symbols found in target:/usr/local/bin/../lib/libpython3.11.so.1.0)
Reading symbols from target:/lib/x86_64-linux-gnu/libc.so.6...
Reading symbols from /usr/lib/debug/.build-id/82/ce4e6e4ef08fa58a3535f7437bd3e592db5ac0.debug...
Reading symbols from target:/lib/x86_64-linux-gnu/libm.so.6...
Reading symbols from /usr/lib/debug/.build-id/ea/87e1b3daf095cd53f1f99ab34a88827eccce80.debug...
Reading symbols from target:/lib64/ld-linux-x86-64.so.2...
Reading symbols from /usr/lib/debug/.build-id/38/e7d4a67acf053c794b3b8094e6900b5163f37d.debug...
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
__GI___clock_nanosleep (clock_id=1, flags=1, req=0x7ffd9d9ec458, rem=0x0) at ../sysdeps/unix/sysv/linux/clock_nanosleep.c:71
71	../sysdeps/unix/sysv/linux/clock_nanosleep.c: No such file or directory.
Traceback (most recent call first):
  (unable to read python frame information)    # <--------------   :(

So it seems like the difference of slim and non-slim isn't only the presence of debug symbols or not, but also the binary is built differently?
Is this possible to solve somehow? My knowledge of how python is built and how to operate gdb are only moderate.

Guida per i contributori

Nessuna guida per i contributori indicizzata per questo repository

Come iniziare

  1. Leggi tutta la issue e poi la guida ai contributi del progetto.
  2. Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
  3. Fai un fork del repository e lavora su un branch.
  4. Apri una pull request che faccia riferimento al numero della issue.

Direzione di ricerca

Non è indicato alcun file del repository né alcun test. Inizia confrontando le build delle immagini python:3.11-slim-bookworm e python:3.11-bookworm, quindi riproduci i comandi di attach di gdb e analizza la discrepanza tra l'eseguibile e i simboli di debug. Il lavoro è completato quando si è stabilito se py-bt può funzionare con l'immagine slim e si è documentato un approccio di debug riproducibile o una limitazione.

Scritto dal modello di indicizzazione a partire dal testo della issue.

Valutazione

Stack tecnologico
docker, python
Ambito
devops, tooling
Tipo di issue
Bug
Difficoltà
4/5
Tempo stimato
3-5 giorni
Stato di attività
Ferma
Chiarezza
Abbastanza chiara
Idoneità per principianti
25/100

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.