fix(ci): replace OC-cluster e2e gate with local docker compose run#203
Open
elronbandel wants to merge 1 commit into
Open
fix(ci): replace OC-cluster e2e gate with local docker compose run#203elronbandel wants to merge 1 commit into
elronbandel wants to merge 1 commit into
Conversation
The per-PR gate was slow (codex node image build) and flaky (bifrost gateway discovery boot race) because it depended on an OpenShift cluster. Replace it entirely with a local docker-based run on the ubuntu-latest runner: build the four images (bench/agent/model/eval) via the CLI, stand up the aime/zerostack/gpt-5.4 compose stack with an output bind mount, and assert on task result.json, agent duration, and gen_ai spans. No OC_SERVER / OC_TOKEN secrets needed. Closes #200
elronbandel
force-pushed
the
fix/e2e-gate-zerostack-gpt54
branch
from
dc13615 to
6343404
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem
The per-PR E2E gate (
oc-connectivity.yml) was slow and flaky because it depended on an OpenShift cluster:codexnode image takes a long time to build on OC.bifrostgateway has a boot-time service-discovery race; when the gateway is unhealthy, every PR goes red independent of the PR diff.This caused every contributor to chase cluster state rather than code quality.
Fix
Replace the OpenShift-based gate entirely with a local docker compose run on the
ubuntu-latestGitHub Actions runner — the same patternnightly-replay.ymlalready uses successfully.What the new gate does:
bench aime,agent zerostack,model gpt-5.4,eval aime --agent zerostacktask/result.jsonexists, agent actually ran (started_at ≠ ended_at), and gen_ai spans are present (warn-only)No
OC_SERVER/OC_TOKENsecrets needed. Secrets required:HF_TOKEN(aime dataset build),OPENAI_API_KEY+OPENAI_API_BASE(gpt-5.4 Azure endpoint),GITHUB_TOKEN(GHCR pull).Rules checked
.agents/contributing/RULES.md— code-only change (no rules modified); rules declared here (R-2, R-3)..agents/RULES.md— change is scoped to one concern (the per-PR CI gate).Closes #200