Skip to content

[codex] Add provider failure classification helpers#24

Merged
senamakel merged 1 commit into
mainfrom
openhuman/error-classification
Jul 5, 2026
Merged

[codex] Add provider failure classification helpers#24
senamakel merged 1 commit into
mainfrom
openhuman/error-classification

Conversation

@senamakel

Copy link
Copy Markdown
Member

Summary

  • add provider-neutral failure classification helpers in harness::retry
  • classify structured ProviderError details, structured HTTP status text, rate limits, upstream unhealthy statuses, and retry-after values
  • route OpenAI provider error construction through the shared classifier

Validation

  • cargo fmt --check
  • timeout 180s cargo clippy --all-targets -- -D warnings
  • timeout 120s cargo test provider_failure
  • timeout 120s cargo test structured_http_status
  • timeout 120s cargo test retry_after_parser_accepts_integer_float_and_space_separators
  • timeout 120s cargo test classify_provider_error_reads_structured_error_fields

@coderabbitai

coderabbitai Bot commented Jul 5, 2026

Copy link
Copy Markdown

Warning

Review limit reached

You’ve reached a temporary PR review limit under our Fair Usage Limits Policy.

Your recent review volume is higher than typical usage, so adaptive limits are currently applied.

Next review available in: 59 minutes

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

How can I continue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details
⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: aad45240-edcd-4ad7-8489-0a6a0a3dbbf3

📥 Commits

Reviewing files that changed from the base of the PR and between e72036d and 8f1b5e7.

📒 Files selected for processing (5)
  • src/harness/providers/openai/sse.rs
  • src/harness/providers/openai/transport.rs
  • src/harness/retry/mod.rs
  • src/harness/retry/test.rs
  • src/harness/retry/types.rs

Comment @coderabbitai help to get the list of available commands.

@senamakel
senamakel marked this pull request as ready for review July 5, 2026 03:30

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 8f1b5e7aba

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".

Comment thread src/harness/retry/mod.rs
_ => message.to_ascii_lowercase(),
};

if status == Some(429) || text_indicates_rate_limit(&lower) {

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Treat statusless quota errors as terminal

When this helper is called from SseState::stream_error it passes status: None, so a streamed error whose code/message is insufficient_quota, quota exhausted, out of credits, or provider code 1311 does not enter this branch unless the text also contains 429/rate-limit wording. It then skips the non-retryable-rate-limit check and falls through to ProviderFailureClass::Retryable, causing the agent retry/fallback loops to repeat permanent billing/plan failures. Check text_indicates_non_retryable_rate_limit before gating on status == Some(429)/rate-limit text.

Useful? React with 👍 / 👎.

Comment thread src/harness/retry/mod.rs
Comment on lines +69 to +72
for (idx, _) in message.match_indices('(') {
if let Some(status) = parse_status_at(message, idx + 1) {
return Some(status);
}

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Stop parsing any parenthesized 3-digit number as status

For statusless provider messages, this loop treats every (...) value that starts with exactly three digits as an HTTP status. A validation error such as invalid embedding dimension (512) or max_tokens (500) exceeds the limit is therefore classified as a 5xx UpstreamUnhealthy and marked retryable, even though no HTTP status was present; this affects streamed errors because they call the classifier with status: None. Restrict the parenthesized parse to known envelopes like API error ( or require an explicit HTTP/status marker.

Useful? React with 👍 / 👎.

@senamakel
senamakel merged commit db56b96 into main Jul 5, 2026
3 checks passed
@senamakel
senamakel deleted the openhuman/error-classification branch July 5, 2026 03:44
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant