This repository was archived by the owner on Mar 9, 2026. It is now read-only.
This repository was archived by the owner on Mar 9, 2026. It is now read-only.
Default retry settings are too aggressive for Resource Exhausted errors #970
Description
With the given initial retry of 0.1 sec, multiplier of 1.3, and max total deadline/timeout = 60 seconds, a Publish request failing with ResourceExhausted could retry 16 times, and it will take over 18 attempts before the retry interval is > 10s.
The backoff should be more aggressive for RPCs that could retry ResourceExhausted: currently, this is just Publish and StreamingPull.