Currently, the Go Admin SDK for FCM places HTTP retry configuration inside an internal package, which makes it inaccessible for customization by developers.
In real-world use cases, different environments may require fine-tuning retry behavior (e.g., backoff strategy, max retries, or disabling retries entirely). Since the retry logic is not exposed, developers have no way to adjust it according to their system requirements.
Suggested improvement:
Expose configuration options for HTTP retry policy in the public API, or
Provide hooks/interfaces so developers can override the default retry behavior.
This would give developers greater control and flexibility when integrating FCM in production systems with specific reliability and latency requirements.