Closed theferrit32 closed 3 years ago
The current build failure in staging is complaining about yarn packages being out of date in the yarn.lock file. We throw --frozen-lockfile
there to ensure that we don't upgrade any packages during the CI pipeline, and that flag stops the install with an error if there's packages that need an update. So, updating and committing a new yarn.lock should fix the staging build.
The build failures in dev are from the cloudbuilder trying to pull the yarn-builder image from the clingen-stage GCR repo, which it doesn't have permission to do. I can look into what's required there to allow the -dev env to pull that image.
@sjahl I'm working on updating the yarn.lock but some package sse4_crc32
is failing to build. There's something broken in the xcode dependency for mac or something.
I also just disabled the cloudbuild trigger on clingen-dev. I'm just doing manual deploys to the dev environment, so I think just having the automated ones go to staging is okay.
If it's useful to do dev builds in cloudbuild once in a while (manually or via some trigger), i think we can make the GCR registry name a variable in cloudbuild, so that dev builds would use a gcr.io/clingen-dev/yarn-builder
image. That would save us from needing to grant the dev env read access to the staging registry.
If running cloudbuilds in dev is not regular or useful, I can forget about that for now 😄
Okay it looks like there is a cloudbuild variable PROJECT_ID
. We could use that in other lines as well.
and here if we change the name of the build targets from app-stage
to clingen-stage
, etc
For dev it would be useful to trigger manual cloudbuilds. My home internet download and particularly upload speed is not very good so doing local docker builds and pushes takes a while.
OK -- I just built a yarn-builder for clingen-dev and am pushing it. I'm also updating and pushing the -stage version so that both environments are using the same thing.
My home upload is also slow, so it'll take a bit for them to show up.
clingen-stage
build triggered by this merge was successful
Actually not sure that this will resolve the current cloudbuild failures, but it will resolve others