I wonder if we could reduce the number of images we maintain here to the minimum and rely more on upstream images instead. Few examples:
I don't think it has a huge impact on cost (registry side), but, less things to maintain is less work 🙃 .
test-runner is not really concerned as it is the image used by prow, so it has everything in it. If we manage to get out of prow in the future, this image would become deprecated.
cc @wlynch @afrittoli @tektoncd/plumbing-maintainers
(Sidenote: we could also migrate the one we can to apko maybe ?)
I wonder if we could reduce the number of images we maintain here to the minimum and rely more on upstream images instead. Few examples:
skopeohas upstream images, and I am not even sure where it is used in our infrastructuretkncould be managed directly ontektoncd/cli, published at release time (and thus, having tagged version aligned with release)hubis very similar to thiskohas an upstream image we could use as well I think (same, it would be tagged per version I think, cc @imjasonh)ko-gcloudis a bit trickierkubectlI am not sure, but it keels like there might be some official or semi-official / maintained image alreadyopenssh-serveris.. a wonder to me, not sure where / how we use that one evenI don't think it has a huge impact on cost (registry side), but, less things to maintain is less work 🙃 .
test-runneris not really concerned as it is the image used byprow, so it has everything in it. If we manage to get out of prow in the future, this image would become deprecated.cc @wlynch @afrittoli @tektoncd/plumbing-maintainers
(Sidenote: we could also migrate the one we can to
apkomaybe ?)