services/backend's indexer (indexer/soroban_rpc.rs) polls Soroban RPC's getEvents on a fixed interval (POLL_INTERVAL_SECONDS), but there's no documented/tested behavior for sustained RPC downtime — no backoff, no alerting, no dead-letter handling if getEvents fails repeatedly.
To do: add exponential backoff on repeated RPC failures, log/alert when the indexer falls behind (compare last successful poll's ledger to current network ledger), and add a test that simulates RPC errors to confirm the poller recovers instead of crash-looping or silently stalling.
services/backend's indexer (indexer/soroban_rpc.rs) polls Soroban RPC'sgetEventson a fixed interval (POLL_INTERVAL_SECONDS), but there's no documented/tested behavior for sustained RPC downtime — no backoff, no alerting, no dead-letter handling ifgetEventsfails repeatedly.To do: add exponential backoff on repeated RPC failures, log/alert when the indexer falls behind (compare last successful poll's ledger to current network ledger), and add a test that simulates RPC errors to confirm the poller recovers instead of crash-looping or silently stalling.