Platform errors
Full node vs archive
RpcNode endpoints target managed full nodes. They are not archive nodes. Calls that need pruned historical state, ancient receipts, or deep traces may fail with upstream JSON-RPC errors (messages vary by client) even when the method name is supported.
Causes
- eth_getBalance / eth_call / eth_getCode at very old block tags on a pruned full node.
- Wide historical eth_getLogs ranges that the node cannot serve efficiently or at all.
- debug_trace* on blocks outside the retained history.
- Assuming Infura/Alchemy archive behavior on a full-node product.
How to fix
- Use latest / recent block tags for state reads when possible.
- Limit log queries to recent ranges; paginate by block window.
- If you need deep history, you need an archive provider — RpcNode documents full-node coverage only.
- Check the network method page: supported method ≠ archive retention guarantee.
Notes
- There is no single RpcNode gateway code for “needs archive”. Failures usually appear as proxied upstream error objects (HTTP 200 with error, or client-specific codes), or as -32002 if the upstream times out.