RPC error HTTP 410 / -32602: beyond_retention

Last updated: September 16, 2026

At a glance

Cause

Stellar history on Triport lives in a measured window, and the two surfaces report its edge differently. Horizon REST answers HTTP 410 Gone with error=beyond_retention and oldest=N; Soroban JSON-RPC answers -32602 with beyond_retention and the oldest available ledger in the message. Both mean the same thing: the ledger you asked for is below the boundary of the node that served you. It is the measured retention of that node, not an internal failure, and repeating the same historical request will not move the boundary.

Solution

1. Read the oldest value from the response and clamp your query to it instead of retrying the same range.
2. Page forward from that oldest ledger rather than backwards from an arbitrary start — Soroban keeps about 120,960 ledgers, Horizon far more.
3. For anything older than the window, serve from your own index; a provider window is an operational buffer, not an archive.
4. Do not treat the error as an empty result: an empty list would let your code conclude that nothing happened, which is the one wrong conclusion here.

Example

{
  "jsonrpc": "2.0",
  "id": 1,
  "error": {
    "code": -32602,
    "message": "beyond_retention: oldest 57926881"
  }
}