Race condition with "docker-compose run -d" and docker rm

[mehertz]

Hi,

I'm seeing what appears to be a race condition between docker-compose run and docker rm. I'm attempting to run docker-compose run on a box which has containers (of the same service!) being actively created and removed and i'm seeing intermittent errors when issuing the next docker-compose run. This can be replicated by running the following command twice:

$ for i in {1..14}; do sudo docker rm test_$i; sudo docker-compose run -d --name test_$i test & done; wait
...
test_7
ERROR: No such container: ea639df9e498c681586a3be619a09381d60896e62b068aa0c884c0998585ee9b
test_14
...

I believe this is because even when just doing a run, docker-compose first grabs the list of available containers, and then does docker inspect on each individual container before spinning up the new container. If an docker rm is issued between grabbing the list of containers and inspecting the container, the above error occurs.

I'm not sure what a good solution is for this. Is it possible for me to do docker-compose run without inspecting every container?

Thanks

[shin-]

Hi!

Yes, unfortunately this is a problem because of the way container naming is done - see #1516
It should be fixed by #5432

[mehertz]

Yeh, I thought that would be it. I'm actually manually setting the container name though, so regardless there's no need to be inspecting every container!

[pauldraper]

Unfortunately, #5432 has been closed :/

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[stale]

This issue has been automatically closed because it had not recent activity during the stale period.