Repository navigation
Monthly from-scratch build of main; SCREAM_MACHINE, and no OpenCode/uv in -dev - #118
Merged
Merged
Conversation
This was referenced Oct 17, 2026
mahf708
force-pushed
the
mahf708/weekly-build-dev-tidy
branch
from
October 17, 2026 04:09
159f271 to
09dbea1
Compare
mahf708
force-pushed
the
mahf708/weekly-build-dev-tidy
branch
from
October 17, 2026 04:32
09dbea1 to
4287a12
Compare
Contributor
There was a problem hiding this comment.
🟢 Approval recommended
No blocking issues were identified; merging remains dependent on #117.
No open issues 🎉
What changed in this PR
Adds monthly rebuild checks for upstream breakage and implements the developer
image changes requested in #115. Merge prerequisite #117 first.
Changes:
- Runs monthly cacheless builds and smoke tests without release promotion.
- Exports
SCREAM_MACHINEwhen the container machine is available. - Removes OpenCode and
uv, updating documentation and smoke checks.
| File | Description |
|---|---|
ghci/tests/smoke-dev.sh |
Checks the machine default and remaining tools. |
ghci/README.md |
Documents machine selection and updated tooling. |
ghci/env/cime/enable.sh |
Defaults SCREAM_MACHINE without overriding users. |
ghci/dev/Dockerfile |
Removes OpenCode and uv installation and setup. |
ghci/dev/devcontainer.json.example |
Removes OpenCode mounts and updates comments. |
.github/workflows/ghci.yaml |
Adds isolated monthly rebuilds without promotion. |
.github/workflows/build-multiarch.yaml |
Adds the reusable no_cache input. |
Review effort: Balanced
mahf708
force-pushed
the
mahf708/weekly-build-dev-tidy
branch
from
October 17, 2026 04:54
4287a12 to
b0ced26
Compare
mahf708
force-pushed
the
mahf708/weekly-build-dev-tidy
branch
from
October 17, 2026 12:42
b0ced26 to
4b2b3cc
Compare
bartgol
approved these changes
Oct 19, 2026
bartgol
left a comment
Contributor
There was a problem hiding this comment.
May I ask what prompted you to remove opencode from the dev images?
Collaborator
Author
it keeps failing download sometimes --- something is finnicky in its link |
Collaborator
Author
|
are you using it? we can readd it later, maybe through a package manager instead of the curl link path |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Second PR in a stack on top of #117 (
gh stack, stack #119). The image builds on main are currently broken by EPEL's Lmod 9.4.2 (and the Intel modules by the Lmod 9.4 compiler-family swap that comes with it), and #117 carries those fixes, so this PR's CI only passes on top of it. Merge #117 first.Monthly build and smoke test of main (
ghci.yaml,build-multiarch.yaml)The Lmod breakage started some time after main's last build on Oct 9, and nothing noticed until #117 happened to rebuild.
ghci.yamlonly ran on PRs, pushes, the merge queue and dispatches. Now:no_cacheis a newbuild-multiarch.yamlinput, so every layer is rebuilt. A cached RPM layer is exactly what would hide a break like Lmod's. The run writes to a-scheduledcache scope of its own and never touches the shared cache.-rc-<run_id>(removed by the existing GHCR cleanup), there's nopromote, and so no spack-cache push. Republishing on a schedule would change the tags people pull with nothing to review. That can follow once Make release builds reproducible and inspectable (pinned base, recorded inventories) #85 pins the base image and records each image's contents.#115
SCREAM_MACHINE=e3sm-container:enable.shexports it next toCIME_MACHINE, under the same rule: only if it isn't set already and~/.cimehas the machine. Sotest-all-eamxxneeds no-min the -dev images, or after opting in. EAMxx's CMake prefers this env var over CIME'sMACH. That makes no difference in a CIME case one3sm-container. With an explicit--machine ghci, a CIME build only uses it for a log line and a cori-knl workaround.uvremoved from the -dev images: the installers,~/.opencode/binonPATH, the pre-created~/.local/share/opencode, the volume and bind lines indevcontainer.json.example, the README mentions, and the smoke-dev checks. smoke-dev now also checksSCREAM_MACHINE.Closes #115.
Checks: actionlint 1.7.12,
check-dockerfiles.py,bash -n/sh -nand the script unit tests pass locally. I didn't build an image locally; this PR's CI covers that. Because it touchesghci/env/cime/, it also runse3sm-caseandomega-case.🤖 Generated with Claude Code