Gate troubleshooting
Start with the complete endpoint copied from Endpoints and a minimal eth_chainId request. Record the HTTP status and any JSON-RPC error without logging the authenticated URL.
Console and key problems
| Symptom | Check |
|---|---|
| No endpoints appear | Select the correct workspace/project and create or select an active RPC key |
| Chain discovery fails | Check that the key is active and belongs to the selected workspace; retry after transient service failures |
| A network is missing | Review the currently discovered list; a network icon elsewhere does not establish endpoint availability |
| Key creation is denied | Use an owner/admin account and confirm an active RPC service subscription |
| New key is unusable | Verify provisioning completed and copy the endpoint again from the active key |
| Old key still needs replacement | Create a replacement, switch applications, then verify completed revocation of the old key |
Request problems
| Symptom | Check |
|---|---|
| Authentication failure | Use the full URL with the RPC key, not the console URL or a Flow/Lift credential |
| Billing or quota refusal | Check the active RPC plan, billing state, and workspace usage |
| HTTP 429 or a capacity error | Reduce concurrency and retry with bounded backoff; follow any returned retry guidance |
HTTP 200 with an error body | Treat it as a JSON-RPC failure and inspect its code/message |
| Method unavailable | Confirm the selected endpoint supports that method; send Lift bundler/paymaster methods to Lift |
| Wrong chain ID | Replace the endpoint or correct the client's chain configuration before submitting transactions |
| Log query rejected | Reduce its block range and add contract/topic filters |
| Historical read unavailable | Confirm archive/historical-state support for that endpoint |
| Repeated connection errors | Check endpoint status and a fresh read; avoid an unbounded retry loop |
Gateway and node errors can use different formats. Parse JSON-RPC responses when available, but handle non-JSON HTTP failures as well. There is no single Gate-specific error-code table guaranteed for all upstream endpoints.
Transactions and timeouts
If submission times out, a transaction may still have reached the network. Look up the known transaction hash and check the sender's nonce before deciding whether to retry, replace, or create a new transaction.
If a transaction is mined but fails, inspect its receipt and contract revert. An RPC connection problem and a reverted transaction require different fixes. Gate access does not provide native gas funding.
For a Lift-sponsored UserOperation, use its Lift operation and receipt to resolve sponsorship and bundling status. A UserOperation hash is not an ordinary transaction hash.
Metrics look different
Compare like-for-like scopes and periods: key traffic is a recent request count, charts use their selected time range, and billing uses the subscription cycle and CU. Usage ingestion can lag a request, and the selected project does not necessarily narrow every workspace billing aggregate.
Useful diagnostic details
When reporting a problem, include the chain ID, method, time in UTC, redacted endpoint hostname, HTTP status, JSON-RPC code/message, and a transaction or UserOperation hash when relevant. Include the key label or identifier rather than the secret itself. Redact the key path segment and any sensitive request parameters.