ZKsync Full Node: Complete Guide

What is a ZKsync full node?

A ZKsync full node syncs an execution client (and, where applicable, the L1/L2 stack behind it), verifies blocks, and serves standard Ethereum JSON-RPC: eth_blockNumber, eth_getBalance, eth_call, eth_getLogs, and eth_sendRawTransaction.

RpcNode offers managed ZKsync Full Node RPC over HTTPS so applications can read ETH balances, simulate contracts, and submit signed transactions without self-hosting Erigon/Geth and companion services.

Why ZKsync full nodes matter

Every wallet balance, payment confirmation, and contract read ultimately depends on nodes that agree on ZKsync state. Public endpoints rate-limit and flake; your product’s reliability tracks the quality of the full nodes behind your RPC URL.

ZKsync in the RpcNode catalog: ZKsync Era.

• Native asset: ETH
• Environments: Mainnet, Sepolia Testnet
• API family: EVM JSON-RPC
• Listed in the RpcNode network catalog

When to run your own ZKsync node

Self-hosting a ZKsync full node makes sense when you need custom indexing, ultra-low latency next to your app, or compliance constraints that forbid third-party RPC. For most product teams, managed Full Node RPC is faster to ship and cheaper to operate.

• Building wallets, payments, or address watchers that need reliable ZKsync reads and writes
• Avoiding multi-terabyte disks, peer tuning, and 24/7 pager duty for chain upgrades
• Preferring HTTPS JSON-RPC with allowlists, keys, and usage metering
• Evaluating ZKsync before dedicating hardware — start on RpcNode, or self-host with Toolkit in a few clicks

ZKsync full node server requirements

ZKsync full node server requirements usually look like other EVM L2 / rollup stacks: an execution client plus a derivation/sequencer companion, often with an L1 RPC dependency.

Publish hardware requirements as full (non-archive) unless you truly need historical state. Exact minimum specs vary by client — always confirm against the network’s official node docs before buying disks.

Looking up "ZKsync server requirements", "ZKsync hardware requirements", or "ZKsync minimum specs"? Use the checklist below as the planning baseline for a self-hosted full node.

ZKsync is in the RpcNode catalog. Until you need custom indexing or on-prem constraints, prefer managed Full Node RPC when available rather than buying this class of server “just in case”.

• CPU: 4–8+ cores depending on throughput
• RAM: 16–32 GB+ typical for full RPC
• Disk: NVMe; start at 500 GB–1 TB and watch growth (archive is multi-TB)
• Bandwidth: 1 Gbps recommended
• OS: Ubuntu 22.04/24.04 LTS
• Full vs archive: full serves tip RPC; archive keeps deep historical state

Self-host ZKsync with RpcNode Toolkit

RpcNode Toolkit is a free self-hosted panel: install a Linux host agent, add the server, then add a node in a few clicks (network → environment → server). The node page walks install, sync, logs, and RPC.

ZKsync may already be in the Add node catalog, or you can use managed Full Node RPC on RpcNode while you wait. Hardware requirements in this guide still apply — Toolkit automates the install, not the disks.

1. Install the host agent on Linux, then Add server in the panel.
2. Add node: pick the network, environment, and server.
3. Open the node page for ports, install, sync, and the fullnode RPC endpoint.

Open Toolkit

How to run a ZKsync full node

How to run a ZKsync full node (EVM family): install the chain’s supported execution client (Geth/Reth/Erigon or a fork / L2 stack), configure a non-archive full profile, and expose JSON-RPC only through an allowlisted proxy.

Many ZKsync-class networks also need an L1 RPC for derivation — budget that dependency before go-live.

ZKsync is in the RpcNode catalog. Until you need custom indexing or on-prem constraints, prefer managed Full Node RPC when available rather than buying this class of server “just in case”.

1. Provision Ubuntu 24.04 LTS (or the OS your client documents) on bare metal or a dedicated VM. Create a system user such as `nodeop` — do not run the node as root.
2. Attach NVMe sized for full (non-archive) growth. Mount it away from the root filesystem and give ownership to the operator user.
3. Apply OS hygiene: unattended-upgrades or a patch cadence, fail2ban/SSH keys, and raise file-descriptor limits if the client docs require it.
4. Install the official or widely supported ZKsync client(s) from upstream releases — pin versions and read release notes before upgrades.
5. Configure full-node sync/pruning defaults suitable for tip RPC. Do not enable archive unless you need deep historical state.
6. Bind HTTP JSON-RPC to localhost (commonly `8545`); keep Engine/auth APIs local.
7. Create systemd services with automatic restart and log rotation.
8. Open only the documented P2P ports publicly; whitelist RPC.
9. Verify with `eth_syncing` / `eth_blockNumber` (or the chain’s equivalent) against a public explorer tip.

• Target: full node RPC, not validator keys (unless you intentionally stake)
• Confirm exact flags in official docs — chain forks differ

Using ZKsync RPC without self-hosting

RpcNode exposes managed EVM JSON-RPC for ZKsync. Create an endpoint in the cabinet, restrict IPs if needed, and call the same methods your stack already uses.

Catalog environments for ZKsync: Mainnet, Sepolia Testnet. Create an endpoint in the RpcNode cabinet for the environment you need, or use this guide to plan a self-hosted deployment.

Typical methods teams start with: eth_blockNumber, eth_getBalance, eth_call, eth_getLogs, eth_sendRawTransaction.

Self-hosting vs managed Full Node RPC

Self-hosting gives maximum control and can reduce per-request cost at huge scale, but you own upgrades, peers, disk growth, and incident response. RpcNode Toolkit cuts the install to a few clicks (agent → Add server → Add node → node page) while you still run the hardware.

Managed RpcNode Full Node RPC trades that ops burden for a stable HTTPS endpoint, usage visibility, and the same method surface your SDK already speaks. If you are comparing providers, also see RpcNode’s comparison page.

ZKsync integration checklist

Use this short checklist when wiring ZKsync into production.

• Pick the correct environment (Mainnet, Sepolia Testnet) — never mix testnet keys with mainnet funds
• Store RPC URLs and keys in secrets management; rotate on staff changes
• Add IP allowlisting for server-side backends
• Alert on tip lag / getHealth (or equivalent) failures
• Load-test eth_getLogs / getProgramAccounts-style heavy calls before launch
• Decide full vs archive needs early — RpcNode is Full Node oriented
• Confirm ZKsync server requirements (CPU/RAM/disk) before self-hosting
• Or install with RpcNode Toolkit in a few clicks — free self-hosted panel

ZKsync full node FAQ