Beacon Kit Changelog

v1.2.0-rc: Bectra

Beacon Kit 1.2.0 adds support for Validator Stake withdrawals and EIP-7702.

Upgrade instructions are posted.

What's new

Since this is a minor version release (major.minor.0), there are breaking changes.

This is a hardfork planned to release onto Bepolia on May 7 2025 at 12 PM EST. Beacon Kit 1.2.0-rc is required to continue progressing on the Bepolia chain after that time.

The -rc will be initially rc0, but there may be more RC releases. Click "Watch" on the Beacon Kit repository home, and monitor the #dev-resources channel on Discord.

The CHAIN_SPEC environment variable is no longer used. There are new options in app.toml for controlling the desired chain to follow. Though defaulted to mainnet so that money machines don't stop working, for Bepolia this configuration must be added:

[beacon-kit] 
chain-spec = "testnet"

During beacond init, for new installations, beacond accepts the new (command line option)[beacon-kit/cli#flags] --beacon-kit.chain-spec.

New Required EL Versions. We have updated our Execution Layer recommended versions to show new versions required for the post-Bectra upgrade. New deployments should use those recommended versions.

To emphasize, do not change anything on mainnet nodes yet. These are for Bepolia only. Mainnet upgrades will be confirmed/announced after Beacon Kit 1.2.0 is out of "release candidate" testing.

v1.1.4: configuration advice

Improves beacond handling of transient conditions (which solve themselves) such as a slow execution layer. It will still exit if the execution layer is shut down.

Also, on startup, beacond now issues warnings about deprecated settings, or settings that could be improved.

• WARN Automatically raising RPCTimeout ... minimum=2000

  Set your RPC timeout to at least 2 seconds:

  FILE: app.toml

  rpc-timeout = "2s"

• ignoring deprecated setting rpc-retries

  This setting is no longer used and should be removed.

  FILE: app.toml

  rpc-retries = 10

• excessive peering

  We recommend that most node operators, including validators, set their maximum inbound peers to 40, and maximum outbound peers to 10. We previously shipped considerably higher default values for this, which can cause excessive memory and CPU consumption.

  FILE: config.toml

  max_num_inbound_peers = 40
  max_num_outbound_peers = 10

• State pruning disabled. This may increase memory footprint considerably

  Setting beacond to disable state pruning, in which previous states are kept, dramatically increases memory usage. Simple RPC nodes can use everything, and validator nodes should use everything. The only exception is if your node is being used to calculate block reward claims, in which case the default setting is a good choice, but this should not be done on the validator itself.

  FILE: app.toml

  pruning = "everything"      # typical
  pruning = "default"         # needed for calculating block reward proofs

v1.1.3: fix deposit mismatch

This release restructures Consensus Layer and Execution Layer communication to keep them in lock-step.

Now, every RPC communication issue among them will result in a BeaconKit termination but keeps their states in sync so you can easily restart any time and keep going.

To that end, configure Reth to not forget blocks on exit with these command line options:

# Reth execution client required flags
--engine.persistence-threshold 0
--engine.memory-block-buffer-target 0

You don’t need to keep these once the node has completed sync, but they make resuming syncing, or syncing from genesis, more robust.

v1.1.2: security improvements

This is a security-focused update:

• Harden timestamp validation of EL payload
• Prevent potential panics and node halts while decoding data

v1.1.1: fixes potential halt

BeaconKit 1.1.1 fixes ASA-2025-001 and ASA-2025-002, which could lead to a network halt. Moreover it hardens some checks around deposit and blob processing.

v1.1.0: BGT minting

BeaconKit v1.1.0 unlocks minting of tokens towards the BGT contract.