ci: add baseline GitHub Actions workflow - #6
Open
draevik wants to merge 2 commits into
Open
Conversation
draevik
changed the title
The repo currently has zero CI visible to PR authors: upstream Prysm's workflows were intentionally stripped (8437139) in favor of Buildkite, but the Buildkite pipeline lives in the Buildkite dashboard rather than this repo, so it gives no automated feedback to anyone opening a PR from outside the org - and nothing public re-runs it. The epoch_processing.go file at the center of the 2026-08-11 incident and its fix had no CI gate in between. Deliberately narrow scope for a first cut: - gofmt check across the module. - bazel build + test restricted to beacon-chain/core/epoch and core/altair - the LightChain-patched consensus packages, so the exact area that caused the incident is covered starting now. Explicitly NOT attempted: full 'bazel test //...' on every PR. This is a large fork with a correspondingly large suite; doing that properly needs a caching/runner strategy (self-hosted runners or a remote cache) that's a separate piece of work, not something to guess at inside this patch. Left as a follow-up, possibly as a scheduled/nightly job instead of per-PR. Untested: I don't have access to this repo's GitHub Actions runners or Bazel remote cache from this environment, so the workflow is reviewed by inspection (valid YAML, package paths confirmed to exist in the tree) but not run end-to-end. Please verify it goes green on a real PR before relying on it as a merge gate. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
…s proposed scope Original patch used Bazel; rewritten to match what issue lightchain-protocol#2 actually asked for - go build ./... plus go test on the packages this fork modifies. Also added cmd/prysmctl (a third LightChain-patched package, per PR1's history) and dropped core/altair from the test step for now: TestProcessEpoch_CanProcess and TestProcessEpoch_CanProcessBellatrix currently fail against main, unrelated to this change - likely stale golden values from the April inactivity-penalty-skip patch, not verified/fixed here. Tracking as a separate issue rather than shipping this workflow red on day one. Verified locally (go1.26.0, WSL2/Ubuntu): go build ./... -> success go test ./beacon-chain/core/epoch/... ./cmd/prysmctl/... -> all ok
draevik
force-pushed
the
chore/baseline-ci
branch
from
August 12, 2026 11:46
26fbb97 to
f697bd6
Compare
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.
What this fixes
Repo has zero CI visible to PR authors (Issue #2). Upstream's GH Actions workflows were stripped in favor of Buildkite, but the Buildkite pipeline lives in the Buildkite dashboard, not this repo — so external PRs get no automated feedback. epoch_processing.go, the file at the center of the 2026-08-11 incident, had no CI gate on it.
What this adds
.github/workflows/ci.yml, matching the "Proposed scope (minimal first PR)" from Issue #2:go build ./...plusgo teston the packages this fork actually modifies. Deliberately not porting upstream's fuzz/clang-format/horusec/etc. workflows in this first pass, per the issue's own suggestion.A finding along the way:
core/altaircurrently fails onmainIssue #2 proposed testing
core/epoch,core/altair, andcmd/prysmctl. Running that locally surfaced two pre-existing failures, unrelated to this PR:In both cases the actual balance is higher than the test expects — consistent with the April patch (
feat(epoch): convert inactivity penalty to forced exit) that intentionally madeepoch_precompute.goskip applying the inactivity penalty to balances. My read is thattransition_test.go's golden values were never updated after that patch landed, and nothing caught it because there was no CI. I haven't fixed this here — I don't want to guess at the correct new expected values for a consensus test without someone who owns the reward-math intent confirming them. Opening a separate issue with these details rather than bundling an unverified fix into a CI PR.core/altairis excluded from this workflow's test step for now (kept ingo build ./..., which does pass) so this baseline CI isn't red from day one. See the follow-up issue for tracking.Verification
Ran locally (go1.26.0, WSL2/Ubuntu):
Both clean:
Before merging, please
Confirm this goes green on GitHub's actual runners (first real end-to-end run). Also flagging: once the altair follow-up is fixed,
core/altairshould be added back into this workflow's test step.Closes issue #2