What Is a Retry Policy in a Service Mesh?

A retry policy in a service mesh tells the sidecar to retry a failed request a bounded number of times when it meets retry conditions such as connection failures or specific status codes. It can add per-try timeouts and backoff so retries do not pile up. The retries happen in the proxy, so application code stays simple.

Why it matters

Automatic retries smooth over transient failures without callers writing retry loops. Set too aggressively, though, they can amplify load on an already-struggling backend.

Related guides