Can a Stellar RPC endpoint resume your event cursor?
A worker comes back online with the next ledger to process saved on disk. A healthy RPC node may already have aged out the start of that backlog.
Compare the cursor with the retained interval
Post this request to the same RPC service used by the worker. Compare the next ledger with oldestLedger and latestLedger. For example, a worker needs ledger 810 while the node reports 830 through 950. The missing prefix covers 810–829 inclusive, exactly twenty ledgers.
{"jsonrpc":"2.0","id":1,"method":"getHealth"}
Keep the gap explicit during recovery
Restore the missing twenty ledgers from a source with suitable history before advancing a contiguous checkpoint to 830. If the saved cursor instead points to 951, wait for the next close and sample again. The boundary moves as the node ingests ledgers, so record the time and boundaries when diagnosing a failed replay. ledgerRetentionWindow describes the retention configuration. Use the observed bounds to decide what this endpoint can currently serve.