taiko-alethia-client-rs v2.2.0

This release prepares the Rust Alethia client for the Unzen Mainnet upgrade and includes important CLI, execution-engine compatibility, checkpoint sync, event sync, proposer, and preconfirmation reliability improvements.

Unzen Mainnet activation

Taiko Mainnet activates Unzen at Unix timestamp 1786021200, corresponding to 2026-08-06 13:00:00 UTC.

Upgrade Rust client, proposer, driver, and preconfirmation deployments before the activation time.

Unzen compatibility

• Update the embedded Alethia-Reth chainspec dependency with the Mainnet activation timestamp.
• Encode Unzen headerDifficulty as a decimal JSON number in engine_newPayloadV2, matching taiko-geth and alethia-reth wire expectations.
• Restore the hash-relevant Unzen header difficulty from Engine API payload envelopes.
• Validate Unzen header-difficulty presence on whitelist preconfirmation envelopes.

CLI changes

Removed

• Remove the preconfirmation-driver subcommand and its permissionless preconfirmation runtime. Use whitelist-preconfirmation-driver for the supported preconfirmation path in this release.
• Remove --preconf.rpc.addr / PRECONF_RPC_ADDR.

Deprecated and ignored

• --p2p.discovery.addr / P2P_DISCOVERY_ADDR
• --p2p.disable-discovery / P2P_DISABLE_DISCOVERY

The deprecated discovery flags are hidden but still parse so deployments can migrate without an immediate startup failure. Discv5 discovery has been removed; configured bootnodes are dialed directly and may use ENR, enode, or multiaddr formats.

Proposer changes

• Restore forced-inclusion inputs so the proposer requests all pending forced inclusions.
• Keep the proposer process running after mined transaction reverts or tx-manager execution-reverted errors.
• Build manifest timestamps from the L1 head and stagger timestamps across multi-block manifests.
• Use the manifest gas limit when selecting transaction-pool content instead of the previous 10M cap.
• Delay missed proposer ticks instead of replaying a burst after a slow confirmation.

Driver and checkpoint sync

• Recover from checkpoint-sync handoff when custom L1-origin mappings are missing instead of entering a restart loop.
• Use a proof-finalized checkpoint target and treat the checkpoint endpoint as an untrusted block-body source.
• Verify the target by block hash and retry transient checkpoint failures.
• Accept VALID or SYNCING during checkpoint import while keeping invalid payloads fatal.
• Enforce strict VALID statuses for normal payload insertion and verify the canonical readback hash after promotion.
• Bound retries for deterministic payload rejections while continuing to retry transient or canonically unproven failures.

Event-sync reliability

• Reconnect and replay after scanner lag instead of silently skipping dropped proposal ranges.
• Identify reorged-out proposal logs through canonical block-hash mismatches.
• Recover from reorged-out logs without blocking event sync indefinitely.
• Reject oversized manifest length fields and trailing bytes after manifest RLP.
• Decode legacy transactions in preconfirmation transaction lists.
• Degrade undecodable derivation-source blobs to the default payload instead of terminating derivation.

Preconfirmation and observability

• Bind queued preconfirmations to their authenticated parent hash and reject stale submissions before engine injection.
• Prevent stale cached envelopes from advancing local status or being served to peers.
• Restore the Go-compatible highestUnsafeL2PayloadBlockID status field.
• Report the execution head from /status when the unsafe counter lags it.
• Add periodic preconfirmation peer-status logging.
• Add the client version metric to the Prometheus endpoint.

Full changelog: taiko-alethia-client-rs-v2.1.0...taiko-alethia-client-rs-v2.2.0