-
Notifications
You must be signed in to change notification settings - Fork 41
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Move image_builder
docker image into Taskgraph
#155
Comments
@jcristau and I were having a bit of debate over this issue. I'll paste our transcript from Matrix here: [jcristauJ [ahal] I guess from a philosophical point of view, your argument would apply to Taskgraph itself. Maybe you feel we should have moved Taskgraph back into gecko_taskgraph rather than the other way around, which is fair.. but also that ship kind of already sailed 😛 So at this point, I think it makes the most sense for image_builder to live beside its sole* consumer * once nss switches to Taskgraph and gecko_taskgraph.docker is merged into upstream 09:20 oh, also I was planning to make Taskgraph pin to the same version of the image as itself. So consumers would upgrade image_builder by upgrading Taskgraph (I haven't worked through the mechanics of that release process wise though) [jcristauJ |
Those are good points. So for moving image_builder to Taskgraph we have: Pros:
Cons:
|
His last commit was also 4 years ago. Tbh, to me this is a sign that it's current location is not great
Gecko could still define an in-tree Docker image that uses |
Last commit from 4 years ago is a reflection of the fact that we haven't needed to touch it because it just works. Every time we touch is spins everything, so it's better to avoid it unless really necessary. There hasn't been a case where we needed to change anything in the last 4 years.
Well, yes an no. The full history is that the image definition was always in mozilla-central, and uploaded do docker hub. But back when we needed to make changes to the image, it was too painful to deal with the docker hub updates (also, it's not self-serve, and can't be tried on try). This is when we ended up with the current approach, which is to use the latest version that had been uploaded to docker hub to create the in-tree one. And all the other images use that one as a builder. Needless to say, with no update in 4 years, it means the current image could have been uploaded to docker hub and used by m-c, and the current setup undone. (I'm not even sure the image that is currently on docker hub uses kaniko) BUT. Had we had done that, bug 1901033 would have been much harder (although arguably if the taskgraph repo has a setup to build the image on PRs, it wouldn't be that bad, but back when this setup was put in place, I'm not even sure taskcluster/github integration was a thing). |
Thanks for the context. And yeah, totally understandable why things are the way they are. I suppose there's room for more automation here regardless of where the image lives. This isn't something I have a ton of motivation to push on atm, so if we're not all in agreement I'll probably just shelve the PR until this comes up again. |
Currently we use a special Docker image to build all the other Docker
image
tasks. But for historical reasons, this image lives in mozilla-central. Now that this repo is the repo of record, and since image tasks are a core feature of Taskgraph, we should migrate the image generation over to here.The text was updated successfully, but these errors were encountered: