finding zombie processes

Sometimes when Docker containers are stopped and started in quick succession for testing I've noticed that my Docker daemon isn't giving a proper call back to systemctl to release the process.

Zombie processes can be difficult to get rid of, and sometimes the symptoms don't necessarily look like they are "zombies".

Example

In one particular instance I had a zombie Docker container process that wasn't releasing an active port.

On startup, this container was throwing a cannot bind to port 9000:9000, it is in use.

Diagnosing zombie processes

When working on this problem, I did the standard docker ps and equivalent checks in Portainer. This returned odd results, as I noticed that the port itself wasn't in use.

I found that zombie processes are a potential issue when shutting down a program, and used this comment from Stackoverflow.

ps aux | grep 'Z'

This returned a process that had not successfully closed, and I then inspected the parent of this process so I could remove the failed process.

pstree -p -s <pid of parent>

After this it was as simple as issueing a kill <parent pid>.