15:10 <_hc[m]> Anyone have a dead simple example of master/worker using DockerLatentWorker? I have it _almost_ working but I'm baffled. I keep getting this in the master's logs:

15:10 <_hc[m]> "

15:10 <_hc[m]> "2022-06-23 15:07:58+0000 [-] Timing out client: IPv4Address(type='TCP', host='127.0.0.1', port=35262)

15:13 <_hc[m]> the worker says:... (full message at https://libera.ems.host/_matrix/media/r0/download/libera.chat/db6ef622ca5a8c5594029bb414734a339f5a4075)

15:34 <Zorry> do it get started okey?

15:42 <Zorry> config ?

16:01 <_hc[m]> When I trigger a build, it does start the container, and I can connect to it via `docker exec`. That's how I got that worker log

16:03 <_hc[m]> master/master.cfg - https://clbin.com/t4ZSi

16:03 <_hc[m]> master/buildbot.tac - https://clbin.com/MO3A4

16:03 <_hc[m]> worker buildbot.tac https://clbin.com/3fik3

16:04 <_hc[m]> worker Containerfile - https://clbin.com/aL2HD

16:05 <_hc[m]> master is running a docker locally (via podman and _podman-docker-)

16:05 <_hc[m]> * master is running a docker locally (via podman and _podman-docker_)

16:05 <_hc[m]> s/podman/_podman_/, s/docker-/docker_/

16:16 <Zorry> set followStartupLogs=True to get more logs

16:17 <glogan> _hc: it's not dead dead simple, but it's somewhat? https://github.com/opencast/opencast-project-infrastructure/blob/master/ansible-buildbot-cluster/roles/buildbot-config/templates/master.cfg#L81

16:17 <glogan> This gets templated out via ansible

16:18 <glogan> I have three worker hosts (this includes the master), and communicate to all three via their tcp connections

16:19 <glogan> the external workers forward their internal ports via ssh tunnels, but you can ignore that bit

16:19 <glogan> I suspect the worker container is the issue here

16:19 <glogan> let me find mine

16:20 <glogan> https://github.com/opencast/opencast-project-infrastructure/blob/master/docker-qa-images/ocqa-deb11-worker-base/Dockerfile#L102

16:20 <glogan> might be important. I seem to remember running into something like this when I first started

18:29 <glogan> ah, I forgot a bit. https://github.com/opencast/opencast-project-infrastructure/blob/master/ansible-buildbot-cluster/roles/buildbot-config/templates/worker.docker is the actual, final worker image. Note the CMD it's using.

18:29 <glogan> IIRC, the docker latent spins up and expect something right away. If it doens't get it then it shuts down

19:50 <glogan> so the default ubuntu 22.04 docker image fails to apt-get update out of the box

19:50 <glogan> because it can't read the keyring files as `_apt`

19:50 <glogan> but you can't fix that as far as I can tell, without apt-get-ing something

19:52 <glogan> ah, because gnupg isn't installed. At all

19:54 <_hc[m]> that sounds like a bug in the docker image

19:55 <glogan> oh lordy. No, it's a bug in docker itself. Crap

19:55 <glogan> https://askubuntu.com/questions/1408090/cannot-run-apt-update-on-ubuntu-22-docker-image-on-a-centos-host

19:55 <glogan> checked on my build nodes -> failure, check on my desktop -> works

19:55 <glogan> docker versions are different

20:19 <_hc[m]> this was one of the bigs that I was missing: `masterFQDN="localhost",`

20:19 <_hc[m]> bits

20:33 <glogan> got it working at least?