Unbounded retries / log storm

What it is

A retry loop has no maximum attempt count or no backoff, or an error path logs inside a loop with no rate limiting.

Why it matters

When the dependency is down, unbounded retries turn one failure into sustained load against a service already struggling, and an unbounded error log turns a bad minute into a very large ingest bill.

How to fix it

Cap attempts, add exponential backoff with jitter, and stop retrying errors that will never succeed. For logging, sample repeated identical errors or aggregate them into a count rather than a line each.