Skip to content

fix: embeddings back off when the model is busy - #1344

Merged
ussaama merged 4 commits into
mainfrom
fix/embedding-rate-limit-backoff
Oct 28, 2026
Merged

ussaama merged 4 commits into
mainfrom
fix/embedding-rate-limit-backoff

Conversation

@spashii

@spashii spashii commented Oct 25, 2026

Copy link
Copy Markdown
Member

On 23 Oct one analysis run of 2636 embeddings hit the per-minute limit; the five short retries ran out and the run failed after 47 minutes of model work. A busy answer now waits as Retry-After says (else a growing, jittered while) for up to eight tries, and holds the other callers on the same embedder while it waits.

@ussaama
ussaama marked this pull request as ready for review October 28, 2026 14:17
@ussaama
ussaama merged commit 1e44de6 into main Oct 28, 2026
17 checks passed
@ussaama
ussaama deployed to pr-preview October 28, 2026 14:40 — with GitHub Actions Active
@github-actions

Copy link
Copy Markdown
Contributor

Now on staging, deployed 2026-10-28 14:53 UTC.

@github-actions

Copy link
Copy Markdown
Contributor

Released in v3.5.3.

This branch was successfully deployed

1 active deployment
pr-preview — 04e77dff Deployed Oct 28, 2026 by ussaama via 51-teardown-pr-preview #1085
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.

2 participants