Hi there!

Concourse version: 3.7.0
Deployment method: docker-compose

Every day, all our builds are frozen, stuck in "pending" state forever, because of check-resources step.
If I try to check resources via CLI, I get an internal server error 500 message.
If I log into the worker container and do "concourse worker", I get this:

tcp 127.0.0.1:7777: bind: address already in use"}}
Exit trace for group:
baggageclaim exited with error: Exit trace for group:
api exited with error: listen tcp 127.0.0.1:7788: bind: address already in use

Sometimes I get an additional error message that :53 is already in use too.

To work around the issue, here is my procedure:
stop the containers (docker-compose stop)
rm the containers (docker-compose rm)
rm the worker container volumes (docker volume prune)
restart the containers (docker-compose up -d)
then I do a check-resource via the CLI, which does not work because it stalls the worker
then I do a fly prune-worker
I restart the containers (docker-compose restart)
then I do again a check-resource via CLI, it works, and the builds can start

...aaaand the issue reappears a few hours later.
It has been an nightmare so far, and all our production is messed up because of this. I can't even keep debugging further as I have no clue about what the concourse binary is doing inside the containers, why it's having issues binding to ports.

Any advice is welcome