HaiAI123

Curated Global AI Tools Directory

NewIndustry update
98
heat index
New

Developer Postmortem Shows Agent Retries Amplifying Outages

11/22/2026 — 11/22, 12:18·1 sources (all)·1 reports (all)

Key takeaways

  • A developer published a postmortem on duplicate outbound calls from an agent: the original retry logic retried 502 errors three times withou…

Background & analysis

On 2026-11-22 at 12:03, the AI section of DEV Community published a first-person post-mortem titled "Most Agents Retry the Wrong Errors: Lessons from a Production Incident", written by a developer who had spent roughly a week tracking down an unusual failure mode. In the write-up, the author explains that their agent's outbound calls would occasionally be sent twice, even though the upstream service was healthy on most days. The trouble started when a flaky gateway began returning 502 responses, and the existing retry loop treated those errors as a coin flip: it retried immediately, three times in a row, with no jitter at all. Under load, that behavior turned a small, transient blip into a thundering herd pressing against the very service that was already struggling.

The author's fix was deliberately simple, built around classifying errors before deciding how to respond. Responses carrying 429 were handled by honoring the Retry-After header and slowing down. Errors in the 5xx range were retried with exponential backoff plus full jitter, while 4xx errors were treated as permanent failures and not retried. The author also capped retries at four attempts, added idempotency keys to write operations, and tracked a per-task retry budget. According to the post, after these changes were rolled out, p99 latency went down and the duplicate side effects caused by repeated calls dropped to zero. No further reports or independent commentary on this incident has appeared, so the write-up from this single developer remains the only available account.

AI-generated from 1 reports · updated 21 minutes ago

updream
24-hour heatHeat index 98
Just charted — trend is building up
24 hours agonow

Reports on this story headlines open the original

Today
  1. After a week chasing double-sent outbound calls, a developer rewrote his agent's retry logic to classify errors first: respect Retry-After on 429s, back off exponentially with full jitter on 5xx, and stop on 4xx. Capping attempts at four, adding idempotency keys and logging a per-task retry budget cut p99 latency and took duplicate side effects to zero.

    DEV Community · AIAI score 56

Other stories people are talking about

How is heat calculated?About the method

Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.

This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.

Surge
Discussion rising fast
New
First report within 6 hours
Rising
Still gathering discussion

Back to AI industry updates →