How to run multiple workers in container?

[jonashaag]

I've started 1 coordinator + 1 worker, each in a Podman container. They have different Node IDs, but when I click on the worker UI (worker.html?<node ID>) I'm shown the info of the coordinator. The worker is also never used.

Both are show as having the same IP address; maybe that's why the UI confused them? They both are shown as having IP 10.0.2.100 (which is NOT their actual IP, but a special IP by Podman).

[mosabua]

Looks like they are completely isolated from each other network wise. How did you deploy them?

[jonashaag]

They are not. The worker can connect to the coordinator (but not receive any work). I can ping/telnet between the machines/servers using the correct IPs and ports. But Trino's mechanism of detecting the IP (and maybe port?) seems to be not working here.

Deployment: Manually start two barebones Ubuntu containers and run the coordinator/worker instance using the launcher script.

[mosabua]

There is no detection mechanism.. you need to configure the discovery server URL (and port) in each node and have that set in config.properties

When you go to the web ui on the coordinator .. does it show the correct number of workers?

Also btw.. the helm chart sorts this all out for you on k8s

[jonashaag]

Yes. It shows the correct number but the UI confuses one for the other. I'll post a screenshot to make it more clear.

[jonashaag]

I added screenshots. Note how the UI confuses the two workers.

[mosabua]

Somehow your podman setup messes with the IP numbers of the workers ..

[jonashaag]

Yep :(

[hashhar]

cc: @electrum might know off the top of his head.