> For the complete documentation index, see [llms.txt](https://docs.triton.one/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.triton.one/solana/faq/rpc-requests.md). # RPC requests Questions about Triton's RPC services, latency, rate limits, scripts and monitoring ## Performance and reliability
How reliable are Triton's RPC services? Our infrastructure runs multiple servers per region with automated health checks. This keeps nodes synced and serving on-tip data. If any node falls behind, requests are routed to a shared backup pool while the issue is resolved. Our infrastructure is monitored 24/7, and we're known for fast and responsive support.
How does Triton ensure minimal latency with RPC? It's a big topic, but at a high level, we do it through: * **Anycast:** routes traffic from anywhere to the nearest server. * **HTTP/3:** RPC over QUIC, with fewer connection round-trips and no head-of-line blocking, which especially helps clients connecting from outside our data centres. * **Custom shred network:** our staked validators forward shreds straight to the RPC nodes, so they receive new block data among the first in Solana's fanout instead of last, where unstaked nodes normally sit. * **Custom indexes:** Cloudbreak optimises `getProgramAccounts` (gPA) calls to improve query performance.
How does Triton handle network congestion on Solana? We mitigate congestion with: * **Staked nodes:** Prioritise transaction delivery to the block producer. * **Load balancing:** Distribute traffic across a fleet of healthy RPC nodes. * **Per-IP rate limiting:** Prevent abuse and ensure fair access. * **Redundancy:** Automatic failover systems to maintain uptime during outages.
What happens if my RPC node receives excessive traffic? Will it affect transaction sending? If traffic exceeds your node's capacity, it may lag behind the network's latest state, resulting in older data being served. In such cases, requests will be rerouted to our backup pool, which supports upgrades and traffic spikes but introduces additional latency. If reliance on the backup pool becomes frequent, we will contact you to discuss scaling your deployment. This doesn't affect transaction submission, as it's routed through a specialised Jet sending engine that isn't "noised" by reads.
## Rate limits and usage
What happens if I hit my rate limit? If you've received `HTTP 429` error, pause requests for 10 seconds to clear the limit. We strongly recommend implementing a backoff-and-retry mechanism. See [Handle 429 rate-limit errors](/solana/faq/error-handling/handle-429-rate-limit-errors.md) for the pattern, or the [Rate and connection limits](/using-your-endpoint/rate-and-connection-limits.md) page for the budgets and headers.
Can I run scripts on the shared service? Yes. Our shared service is as performant as private clusters. For gRPC, we recommend [Riptide](/solana/real-time-streaming/riptide.md) over the standard Dragon's Mouth endpoint. If you expect sustained heavy load (heavy subscriptions, full-chain streams, custom Geyser configurations), a dedicated node might be a better fit.
How do I check my current rate limits? Your endpoint exposes your live limits at `/ratelimits`. Run `curl https://.mainnet.rpcpool.com//ratelimits`. JSON-RPC responses can also carry `X-Ratelimit-*` headers (your tier, plus limit and remaining pairs for requests, the endpoint, the method, and connections), though not every endpoint returns them. Where they appear, use them to back off before you hit a 429. See [Rate and connection limits](/using-your-endpoint/rate-and-connection-limits.md).
## Configuration and access
Do you support devnet? Yes. Triton runs Solana mainnet and devnet, with full compatibility with the standard JSON-RPC and WebSocket APIs on both. Dev/test and mainnet subscriptions are created at sign-up for each chain you pick. If you need another chain later, contact support from your [customer dashboard](https://customers.triton.one) to add it.
How do I monitor my RPC usage? Customers with dedicated nodes get a Grafana dashboard with real-time metrics: request volume, latency, error rates, and bandwidth usage. For shared (pay-as-you-go) usage, open the **Billing** tab in your [customer dashboard](https://customers.triton.one) to see total requests and GB used (the two dimensions you're billed on), plus a per-service breakdown.
Can I create new tokens or endpoints myself? Accounts with standard permissions can't create tokens or endpoints directly. Contact support by clicking the chat icon in the bottom right of your [customer dashboard](https://customers.triton.one) to provision them. If you need self-service, ask support to upgrade you to the operator role; once enabled, you can add and deactivate tokens from the dashboard yourself. New endpoints are still provisioned by support.
What's the __blocked.rpcpool.com origin on my endpoint? `__blocked.rpcpool.com` is a system origin that Triton adds when you turn on an endpoint that has no allowed origins. It allows no browser origin, so the endpoint only accepts requests authenticated by a valid token. It counts toward your active origins while it's active. The dashboard won't let you save an active endpoint with no active origins. See [Auth and security](/security/authentication.md).
Can I choose the geographic location of my RPC nodes? Yes, dedicated node users can select primary regions (US, EU, or Asia Pacific) or request multi-region setups for redundancy and lower latency.
Can I optimise for a specific region? By default we use BGP Anycast to route every request to the nearest city automatically, requiring zero client-side setup. On shared Solana mainnet, to pin your traffic to one city, replace your endpoint's hostname with a city hostname, such as `https://.rpcpool.com/`. Support can also add a hostname for your endpoint that points at one city. See [Endpoints, regions, and routing](/using-your-endpoint/endpoints-regions-routing.md#pin-a-region).
Are shared RPC subscriptions restricted to a single domain? No. You can whitelist several domains per endpoint from your customer dashboard: up to 5 active origins by default, and support can raise the limit.
## Historical data and add-ons
Can Triton provide historical data? Yes, shared mainnet endpoints include access to the complete ledger through the Superbank pipeline. Dedicated node customers can get the same history from a shared mainnet subscription added to their account. Dedicated node users can also get archival nodes set up that store complete Solana history (e.g., past blocks, transactions). This requires additional storage (multi-TB scale) and is priced separately, reach out for a custom quote.
What is Jetstreamer? Jetstreamer is the tool for bulk-backfilling large ranges of Solana history from the Old Faithful archive. It writes to ClickHouse or to a Jetstreamer plugin you write, which can filter the data, and works with existing Geyser plugins for epochs 157 and later.
Can I use Jito sendBundle and bundle simulation with Triton RPCs? Bundle simulation is available to everyone: see [Jito Bundle Simulation](/solana/sending-transactions/3rd-party-apis/jito-bundles.md) for the full reference. `sendBundle` itself isn't routed through our infrastructure. Routing bundles through Triton would add an extra hop in front of the Jito block engine, which adds latency and runs counter to the reason you're using bundles in the first place. The recommended pattern is to call the Jito block engine directly for sends, and use Triton for everything else (reads, streams, simulation).
## Troubleshooting
I have an issue with my Web3.js socket connection. How can I fix it? If you're hitting persistent socket or connection errors with `@solana/web3.js` (`fetch failed`, `Connect Timeout Error`, `ECONNREFUSED`, `ECONNRESET`, `other side closed`), see [Web3.js socket and connection issues](/solana/faq/error-handling/web3js-socket-connection-issues.md). It covers the root causes and the standard fixes.
How can I accurately measure latency when using Triton's RPC services? To measure latency effectively, avoid relying on blockchain or explorer timestamps, as these may not reflect the actual performance of your RPC provider due to drift in the timestamping. The on-chain time and clock time drift apart, leading to misleading metrics. Instead, record a timestamp immediately after receiving a transaction response from each provider in your application. This method captures the actual time it took for the request to complete, providing a reliable metric for assessing latency.
How can I monitor for delays on my gRPC connection? To detect delays, examine the `slot` field included in all gRPC messages. The most precise method is to compare this slot value with a secondary source, such as another subscription or a `getSlot` call in your production environment. For additional validation, you may also cross-reference with an alternative RPC provider. Dedicated users can select the "Tracking Tip" folder in their Grafana Dashboard. The slot latency reflected here shows how much your node's slot latency differs from the rest of the network. As the node processes data from the network downstream, it sends this information through Geyser. If you suspect your Geyser stream is experiencing a drastically different latency, check whether your receiving server is not providing back pressure to the RPC node due to bandwidth constraints.
My code uses an old RPC method and now gets an error. What replaced it? Solana v2.0 permanently removed several legacy JSON-RPC methods. See [Solana 2.0 removed RPC methods](/solana/solana-2-0-removed-methods.md) for each removed method and its successor.
*** :life-ring: Contact support by clicking the chat icon in your [customer dashboard](https://customers.triton.one)\ :briefcase: Sales questions? [Contact us](https://triton.one/contact)\ :sparkles: AI agent? Read [llms.txt](https://docs.triton.one/llms.txt)\ :rss: Follow updates: [Blog](https://blog.triton.one) · [X](https://x.com/triton_one) · [YouTube](https://www.youtube.com/@triton_one_ltd) · [Telegram](https://t.me/tritonone) · [GitHub](https://github.com/rpcpool)