thanks @Samhita Alla, will run using the docker image

Can try running the command I sent you earlier:

flytectl demo exec -- docker build . --tag "flyte_select_wf:v1"

?

yeah. Is there any alternative for quick check, or some working example I can glance through where an external library and files are imported in the workflow python file?

Can you try https://docs.github.com/en/packages/working-with-a-github-packages-registry/working-with-the-container-registry? Try pushing the image to your github account. It’ll by default be private. You can follow the private registry guide to access the images in case you want the image to remain private.

alright will do

@Kevin Su, am I right? Is there a way to access docker image in a demo cluster if the image is present locally?

yeah, I don’t think image can be accessed in k3s cluster if it’s present locally. You can also try to push to docker hub

docker tag flyte_select_wf:v1 <docker_account>/flyte_select_wf:v1
docker push <docker_account>/flyte_select_wf:v1

[1/1] currentAttempt done. Last Error: USER::containers with unready status: [a79kkwndktg4rbwswjp9-n0-0]|Back-off pulling image “ghcr.io/main2adwitiya/gunjan_select:latest”

@jeev / @Yee fyi. @Adwitiya we are infact working on adding a dummy docker registry to the demo environment

@Adwitiya so the

flytectl demo

environment runs a docker in docker image - there’s a separate docker daemon in the container (the container that’s running k3s). To get around this you can do a couple things - 1 suggested above, push to an external docker registry like dockerhub or ghcr, or 2, run the demo environment with

flytectl demo start --source .

This will mount in the current folder into the docker in docker container, and you can build your image inside, where it will be visible to the k3s cluster. There’s a longer term fix coming out in the next couple weeks where the demo environment will just spin up a local docker registry that is pushable right from your desktop environment but pullable from within k3s. this means we can move away from the

--source

flag since you won’t need to mount your code into the d-in-d container.