fix(runner): retry a run after a provider rate limit before any tool call - #45
Merged
Merged
Conversation
…call Opus runs through the AI Gateway fail with 429 Wholesale Rate limited even at concurrency 1: 51 of 80 Opus adoption runs in the 09-07 matrix. A 429 before the agent has touched the workspace is not a result, so the runner now waits (15s, 45s, 90s) and starts the conversation again, up to three times, and records the retries in the run metrics. A rate limit mid-run still fails the run: replaying tool calls on a modified workspace would not be the same measurement.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
In the 09-07 weekly matrix, 51 of 80 Opus adoption runs failed with
429 Wholesale Rate limitedfrom the AI Gateway even at concurrency 1, so the site's Opus adoption cells show "Failed". A 429 before the agent has touched the workspace is not a measurement; the runner now waits (15s, 45s, 90s) and starts the conversation again, up to three times, and recordsrateLimitRetriesin the run metrics (optional field, no schema bump). A rate limit mid-run still fails the run, since replaying tool calls on a modified workspace would not be the same measurement.pnpm format && pnpm lint:fixto format the code