fix(ci): move Build workflow off the deprecated ubuntu-20.04 runner - #28
Open
Akshay-2007-1 wants to merge 8 commits into
Open
Akshay-2007-1 wants to merge 8 commits into
Akshay-2007-1 wants to merge 8 commits into
Conversation
GitHub removed the Ubuntu 20.04 hosted runner image, so this job was sitting queued forever with no runner ever picking it up (confirmed: zero prior runs of this workflow existed, and the release-v0.3.0 tag push got stuck queued for 20+ minutes with no runner_name assigned until manually cancelled). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UfpKt8WtJCGZHiETgDXmYu
actions/upload-artifact@v2 is hard-deprecated - GitHub auto-fails any run using it before a single step executes. Bumps it and the similarly ancient actions/checkout@v2 to v4. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UfpKt8WtJCGZHiETgDXmYu
satackey/action-docker-layer-caching@v0.0.8 calls a GitHub Actions
cache API endpoint that's since been decommissioned ("Our services
aren't available right now"), failing every run before the build even
starts. It's purely a build-speed optimization, not required for
correctness - removing it rather than trying to replace it.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UfpKt8WtJCGZHiETgDXmYu
…control's Dockerfile
BuildKit mounts each RUN step's rootfs with nosuid, which breaks
`sudo` there ("effective uid is not 0") even though the Dockerfile
itself is unchanged - this only surfaced from the runs-on bump to a
runner image shipping a newer Docker Engine. Falls back to the legacy
builder, which doesn't impose this restriction.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UfpKt8WtJCGZHiETgDXmYu
…uilding Confirmed the nosuid failure isn't BuildKit-specific (same error with DOCKER_BUILDKIT=0) - it's the runner's Docker host mounting RUN steps' rootfs nosuid regardless of builder. sudo was never load-bearing here anyway (build_control_panel.sh always runs the resulting image with -u 0:0), so removing it is safe and sidesteps the host-level issue entirely, without touching the upstream submodule repo. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UfpKt8WtJCGZHiETgDXmYu
…ad of relying on sudo Root cause, verified by comparison: the sibling debian-stretch-cross image (Dockerfile.sling) uses the identical sudo pattern and builds fine on this same runner, but debian-stretch-armel-cross's own /usr/bin/sudo is missing its setuid bit - a defect specific to that one base image, not a runner/host/BuildKit issue (previous two fix attempts for those wrongly assumed causes didn't help, and confirmed this image's default user genuinely isn't root either: a plain sed -i on /etc/apt/sources.list without sudo fails with a real permission error). Forcing USER root sidesteps sudo's broken setuid mechanism entirely rather than depending on it. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UfpKt8WtJCGZHiETgDXmYu
… keys apt refuses to install packages non-interactively once a repo's Release-file signature has expired, which is the case for archive.debian.org's Stretch suite (confirmed via the build log's GPG warnings: both the Debian Archive Automatic Signing Key and the Debian Stable Release Key are expired). Scoped narrowly to this one install step, from this single archived/dead repository only. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UfpKt8WtJCGZHiETgDXmYu
…of hardcoding them
The hardcoded exclusion list assumed the CI host's mke2fs would either
recognize each feature name (and disable it) or never have heard of it
at all (making the exclusion a harmless no-op). Bumping the runner
image broke that assumption in a new way: its mke2fs (1.46.5) doesn't
recognize one of the five names, and an unrecognized name in the -O
list isn't a no-op - it makes mke2fs reject the whole option string as
invalid, aborting the build ("Invalid filesystem option set").
Replaces the hardcoded string with a real per-feature test against a
scratch file on this same host's mke2fs (the same binary guestfish's
appliance is built from), keeping only the names it actually accepts -
self-adapting to whatever e2fsprogs version happens to be installed,
rather than needing another hardcoded update next time it changes.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UfpKt8WtJCGZHiETgDXmYu
This branch has not been deployed
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
1 participant
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.
Summary
Buildworkflow'sruns-on: ubuntu-20.04requests a GitHub-hosted runner image that's been deprecated and removed. The job sits queued forever with no runner ever picking it up.release-v0.3.0tag queued a run that sat with an emptyrunner_namefor 20+ minutes before being manually cancelled - and this workflow had zero prior successful runs in the repo's history to compare against.ubuntu-22.04, still a currently-supported GitHub-hosted image.Test plan
workflow_dispatchonmain, or re-push arelease-*/dev-*/sling-*tag) and confirm a runner actually picks it up and the build completes.🤖 Generated with Claude Code
https://claude.ai/code/session_01UfpKt8WtJCGZHiETgDXmYu