What each public endpoint is, and what it is not good for.
The endpoints
| Endpoint | What it is |
|---|---|
https://rpc.aere.network | Public JSON-RPC read endpoint |
https://rpc2.aere.network | Second public JSON-RPC endpoint |
wss://wss.aere.network | WebSocket, for eth_subscribe |
wss://api.aere.network/ws | Second WebSocket endpoint |
https://explorer.aere.network | Block explorer, human-facing |
https://api.aere.network | Indexer REST API, for indexed history |
All of them are operated by the AERE Foundation.
Which one for which job
Reading the chain right now: either HTTPS endpoint. eth_blockNumber, eth_getBlockByNumber, eth_call, eth_sendRawTransaction and the rest of the standard set.
Watching for new blocks: a WebSocket endpoint with eth_subscribe. Subscriptions are not available over HTTPS.
History for an address: the indexer, not the RPC. Scanning blocks over RPC to find an address's transactions is slow and will time out. See Indexer API.
Proving something about an account: eth_getProof, and only that. See the state window below.
The state window, and the false zero
The public RPC keeps state for 512 blocks. Beyond that window, state-reading calls answer from a pruned state, and eth_getTransactionCount returns a false 0x0 rather than an error.
That is the dangerous shape: an answer that looks like a measurement. A tool that reads a nonce at an old block and concludes "this account has never transacted" is wrong, and nothing tells it so. Prove facts about accounts with eth_getProof, which either answers or fails.
At 512 blocks and half-second blocks, the window is roughly four minutes of chain.
A node that is down is not a node that says zero
If you build monitoring against these endpoints, treat "no answer" as its own state and never as zero. A node that is catching up may not answer RPC at all, and a peer count or a height read as zero in that moment is not a measurement of anything.
Rate limits and etiquette
The endpoints are public and unauthenticated. They are also a small number of real machines. If you need sustained throughput, run your own node: the package, the genesis and the exact options are published for that. See Run a node.