Repository navigation
[Feature]: Optional same-target retry (configurable interval) before failover on 429 #487
Closed as not planned
Description
Version: v2.7.40
Current behavior
On 429, the combo immediately cools the target down and routes to the next target. There is no way to say "wait N seconds and try the SAME target again".
Desired behavior
An optional per-combo retry policy, e.g.:
{ "retry": { "on": [429], "attempts": 10, "intervalMs": 5000 } }Retry the current target every 5s up to 10 times; only then fail over to the next target. For "primary-first" users this is preferable to switching to the fallback on every rate-limit blip.
Context
The Codex client itself does not retry 429 (upstream: openai/codex#30471) — it only retries 5xx — so a proxy-side retry would fill a real gap. Today the only workaround is failover to a fallback provider, which changes the model/quota being consumed.
Activity
github-actions commentedon Jul 26, 2026
Issue closed: insufficient detail
Thanks for taking the time to submit this issue. It was closed automatically because the structured report is missing information needed to evaluate or reproduce it.
Please edit the issue to add:
See the Contributing guide for details.
Once the report passes the automated checks, it will be reopened automatically unless a maintainer has changed its state.
github-actions commentedon Jul 26, 2026
Automated translation bookkeeping — detected language: English.
a1980881039 commentedon Jul 26, 2026
One more note: if you're busy, I'd be happy to take a stab at this myself and open a PR. Just let me know.
feat(proxy): opt-in same-target 429 wait-and-retry before key failover (
Merge pull request #865 from harryzhou2000/feat/429-same-target-retry
feat(proxy): opt-in same-target 429 wait-and-retry before key failover (
Merge pull request lidge-jun#865 from harryzhou2000/feat/429-same-tar…