Supervisord does not show stdout from processes viewed via docker

Question

Trying to capture logs of my app with supervisor in docker.

Here's my supervisord.conf:

[supervisord]
logfile=/dev/null
nodaemon=true

[program:autofs]
command=automount -f
redirect_stderr=true
stdout_logfile=/dev/stdout
stdout_logfile_maxbytes=0

[program:split-pdf]
command=bin/split-pdf-server
directory=/root/split-pdf
redirect_stderr=true
stdout_logfile=/dev/stdout
stdout_logfile_maxbytes=0

After starting container everything works, and i can see the result of my app running (it creates pdf files on network share)

But the log shows no output from my app:

015-07-02 00:39:26,119 CRIT Supervisor running as root (no user in config file)
2015-07-02 00:39:26,124 INFO supervisord started with pid 5
2015-07-02 00:39:27,127 INFO spawned: 'split-pdf' with pid 8
2015-07-02 00:39:27,130 INFO spawned: 'autofs' with pid 9
2015-07-02 00:39:28,132 INFO success: split-pdf entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)
2015-07-02 00:39:28,132 INFO success: autofs entered RUNNING state, process has stayed up for > than 1 seconds (startsecs)

That's the only output i see when attaching to docker container.

I'm on ubuntu 15.04 docker 1.7.0

This is not a duplicate of this question, because i'm running more than one process in a container.

Answer 1

This has to do with pipe buffering.

I was able to circumvent that problem by running python in unbuffered mode:

$ docker run -e PYTHONUNBUFFERED=1 imagename

See supervisor-stdout issue #10 for a discussion on this.

Answer 2

It turns out everything works, but with some delay. When i tried to create a container for some other app that produces much more log messages, the messages started appearing in the log file, but with delay.

The first app i was testing it with had only 2 lines per task in the log, and i guess there's some kind of buffer that needs to be filled before it starts flushing to the log file.