BUG: Docker TLS mode broke `docker run` and `docker exec`

Rack (Version 2)
1

Cross-posting from GitHub:

After having updated one of our racks from 20220817173516 to 20221013170042 we’ve noticed, for example, convox exec <pid> sh --app <app> is broken and it will fail with the following error message:

ERROR: Get "https://10.1.2.4:2376/containers/json?all=1&filters=%7B%22label%22%3A%5B%22com.amazonaws.ecs.task-arn%3Darn%3Aaws%3Aecs%3Aus-east-1%3A924748183198%3Atask%2Fmyrack-Cluster-SUZR61FH8VHZ%2F6f2895b834cf4b09a7df955826c4adb1%22%2C%22convox.release%22%5D%7D": http: server gave HTTP response to HTTPS client

This is a show-stopper for us, so we had to downgrade our rack whilst waiting for a fix.

Any ideas?

2

We figured our issue was due to our custom InstanceBootCommand which didn’t include recently added --tls --tlscacert /etc/ca.pem --tlscert /etc/cert.pem --tlskey /etc/key.pem parameters for Docker.

I wish this breaking change was mentioned in the release notes. :v: