docker-py：访问在ENTRYPOINT命令中运行的python脚本输出

Question

I'm having trouble getting the output of a Python script I'm running in a docker container using the docker-py module in Python.

First, some context:

I've created a Dockerfile and built my image (with id 84730be6107f) in the normal way via the command line ( docker build -t myimage /path/to/dockerfile ). The Python script is executed as the ENTRYPOINT command in the Dockerfile:

ENTRYPOINT ["python", "/cont_dirpath/script.py"]

The directory containing the script is added (bound) to the container upon running it.

When I do this via the usual docker command line (in Ubuntu) using:

docker run -v /host_dirpath:/cont_dirpath 84730be6107f

I get the expected Python script output printed to stdout (displayed in the terminal - the script ends with a print result command). This is exactly the output I'm trying to get while doing this from within Python.

However, when trying to do this from within Python using the docker-py package, I am unable to view the results - the logs method returns an empty string. Here's the Python code I'm using:

from docker import Client

docker = Client(base_url='unix://var/run/docker.sock',
                  version='1.10',
                  timeout=10)

contid = docker.create_container('84730be6107f', volumes={"/cont_dirpath":""})
docker.start(contid, binds={"/host_dirpath": {"bind": "/cont_dirpath"} })

print "Docker logs: \n" + str(docker.logs(contid))

I just get an empty string. I've tried piping the output of the script to echo and cat in the ENTRYPOINT command but to no avail. How can I get the output of script.py run in the ENTRYPOINT command to be part of the log for that container so I can read it using the docker-py Python methods? I've also tried the docker-py attach method (I think logs is just a wrapped version of this) with the same result.

I've considered just writing the script output to a file instead of stdout, but it isn't clear how I would access and read that file using the docker-py methods.

Any insights or suggestions would be greatly appreciated!

Answer 1

What happens when you try to view the logs with kubectl ?

kubectl logs $CONTAINERID

I'm guessing you'll see nothing.

What is your workload inside the container?

Python does something weird with buffers when run headless if your code looks like this:

import time
while True:
   print('something')
   time.sleep(1)

To make it work, you need to manually flush the looped print statements:

import time
import sys
while True:
   print('something')
   sys.stdout.flush()
   time.sleep(1)