Verify strategy plugin mounts after deploy#161
Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 6deabb799a
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| if expected_schema and str(signal.get("schema_version") or "").strip() != expected_schema: | ||
| raise ValueError( | ||
| f"{service}:{env_name} expected schema {expected_schema}, " | ||
| f"got {signal.get('schema_version')!r} at {signal_path}" | ||
| ) |
There was a problem hiding this comment.
When a mounted signal_path still contains valid JSON with the expected schema but an old as_of/generation timestamp, this verifier accepts it because the only content check after reading GCS is the optional schema_version comparison. That lets the new post-sync step report success while Cloud Run is left consuming a stale strategy-plugin signal, which is one of the failure modes this verifier is meant to catch.
Useful? React with 👍 / 👎.
| env: | ||
| STRATEGY_PLUGIN_MOUNT_ENV_NAMES: LONGBRIDGE_STRATEGY_PLUGIN_MOUNTS_JSON |
There was a problem hiding this comment.
When an environment needs strategy-plugin signals from a bucket other than the hard-coded gs://qsl-runtime-logs-shared/ default, this workflow has no way to provide STRATEGY_PLUGIN_ALLOWED_SIGNAL_PREFIXES to the verifier even though the script reads it. GitHub configuration variables are not automatically exported into the runner environment, so a valid Cloud Run update can be applied and then fail verification solely because the override was never wired here.
Useful? React with 👍 / 👎.
1 participant
Summary
Validation