Troubleshooting Fedora Atomic Host 28


#1

I’ve got a containerized application that is terminating mysteriously. It starts fine but towards the end of the run it crashes. It’s a bash script and all I see is “Killed”. How can I find out what’s killing it and why?


#2

So, it could be killed for numerous reasons from taking too much memory, trying to do something forbidden by seccomp, etc. What is the application, how do you start it, etc, would help us provides more ideas.


#3

It’s a bash script that runs some Python code to insert data into a PostgreSQL database. :wink: The container is running the Docker Store PostgreSQL image with some other stuff. I’m going to break it into two containers - one just for PostgreSQL and the other just for Python / bash - before I do any more troubleshooting. I’ve got some Python environment issues in another script that I want to isolate.

I did try doubling the host RAM from 1 GB to 2 GB but that didn’t help. If it’s seccomp or something like that, where would I look for log files?


#4

I would start the container and monitor journalctl during this time to see if there are any relevant messages in the log.


#5

Thanks! I’ve got Cockpit set up now too, so I can monitor memory usage.


#6

It turned out to be a Python script that spiked up to 5 GB in the final phases of its operation. Just enough to trigger the OOM killer in a 4 GB Digital Ocean droplet. :frowning: