Cloudflare · infra · fg-0591

Durable Object calls now retried by the runtime, so your own retry loop multiplies attempts

If you see a failed Durable Object call takes up to about 10 seconds longer to surface an error, the cause and the fix for Cloudflare are below, last verified 2026-11-06.

Verified Nov 2026trust 50/100
Symptom
a failed Durable Object call takes up to about 10 seconds longer to surface an errortotal attempts exceed your configured retry countnon-idempotent method seems not to rerun, while retryable-marked methods may run twice
Cause
From 2026-11-05 the runtime retries fetch() and RPC calls to a Durable Object up to four times, none starting more than 10 seconds after the first call, when they fail with a retryable error before reaching the object. Errors thrown by your object code and errors with .overloaded true are not retried. Retry options need Wrangler 4.158.0 or later.
Fix
Lower your own attempt count, or set maxRetries to 0 on stubs where your code handles retries. Mark only idempotent methods with retryable() from cloudflare:durable-objects.
Apply
const stub = env.CART.getByName("user-123", { retry: { maxRetries: 0 } });
Affected
All Durable Object stubs called from Workers or other Durable Objects, from 2026-11-05
Sources

Last verified 2026-11-06 · Confirmed by 0 builders

Related traps

Have your agent know this before it fails

$ /plugin marketplace add getfootgun/footgun
$ /plugin install footgun@footgun