github ava-labs/avalanchego v1.15.0
Helicon: Continuous Execution & Auto-Renewed Staking
on GitHub

2 hours ago

This release schedules the activation of the Helicon network upgrade at 11 AM ET (3 PM UTC) on Tuesday, September 22nd, 2026 on Mainnet.

All Mainnet nodes must upgrade before 11 AM ET, September 22nd 2026.

This release updates the plugin version to 46. All plugins must update to remain compatible.

Features

Helicon activates the following Avalanche Community Proposals (ACPs):

  • ACP-194 C-Chain Async Execution
  • ACP-236 Auto-Renewed Staking
  • ACP-267 Validator Uptime Requirements Increase
  • ACP-273 Reduce Minimum Validator Staking Duration
  • ACP-283 Dynamic Minimum C-Chain Gas Price
  • ACP-285 Reduce Minimum Consumption Rate

Other features:

  • Subnet-EVM can set the initial ACP-226 minimum block delay at genesis via InitialMinDelayMS in the chain config.

C-Chain State Sync

C-Chain state sync is not supported immediately around the Helicon activation. A node that is state syncing when Helicon activates may stall.

After Helicon activates, state sync has the following limitations:

  • Restarting a node mid-state sync will restart the sync from the beginning.
  • Disabling state sync after partially state syncing may result in an unrecoverable FATAL error.
  • If state-scheme is firewood, the node cannot state sync. A new Firewood node on Mainnet or Fuji MUST set state-sync-enabled to false or will shut down with a FATAL error.
  • On custom networks, a new node starts on the pre-Helicon Coreth VM and bootstraps by executing all blocks.

If encountering one of these FATAL errors, the only way to recover is by deleting the database and restarting.

APIs

C-Chain RPCs

  • Named blocks under SAE:

    • pending is the last executed block by default. With api-resolve-pending-to-last-executed set to false, pending is the last accepted block.
    • latest is the last executed block.
    • safe and finalized are both the last settled block.

  • The avax.getAtomicTxStatus RPC is deprecated. Use avax.getAtomicTx.

  • The admin, warp, and personal API namespaces are removed.

  • The eth_accounts, eth_coinbase, and eth_etherbase RPCs are removed.

  • The following debug RPCs are removed:

    • debug_dumpBlock
    • debug_preimage
    • debug_getBadBlocks
    • debug_accountRange
    • debug_storageRangeAt
    • debug_getModifiedAccountsByNumber
    • debug_getModifiedAccountsByHash
    • debug_getAccessibleState

    If you rely on any of these debug RPCs, open an issue describing your use case.

Metrics

  • Histogram avalanche_snowman_consensus_latencies has 8 buckets instead of 4, each a second wide.
  • Added avalanche_evm_transition_sae_last_executed_height and avalanche_evm_transition_sae_last_settled_height gauges, exposing SAE execution and settlement heights.
  • Added SAE execution-pressure metrics:
    • avalanche_evm_transition_sae_execution_queue_duration_seconds (histogram): time from a block's acceptance into the execution queue until its execution completes.
    • avalanche_evm_transition_sae_execute_block_duration_seconds (histogram): wall time to execute a single block, including the state commit and post-execution work.
    • avalanche_evm_transition_sae_execution_queue_blocks (gauge): number of accepted blocks that have not yet completed execution.
    • avalanche_evm_transition_sae_execution_queue_gas_limit (gauge): worst-case gas of accepted blocks that have not yet completed execution. Worst-case gas is the sum of transaction gas limits and end-of-block operation gas.
    • avalanche_evm_transition_sae_executed_gas_charged_total (counter): cumulative gas charged by executed blocks. Charged gas is the sum of transaction gas used and end-of-block operation gas.
    • avalanche_evm_transition_sae_executed_gas_limit_total (counter): cumulative worst-case gas of executed blocks.
  • Added avalanche_evm_transition_sae_in_memory_blocks (gauge): number of SAE blocks still live in memory (created but not yet garbage collected).
  • Added avalanche_evm_transition_sae_accepted_gas_limit_total (counter): cumulative worst-case gas of blocks accepted into the execution queue; the acceptance-side counterpart of executed_gas_limit_total.
  • Added SAE gas-time and pricing metrics:
    • avalanche_evm_transition_sae_last_executed_gas_time_seconds (gauge): gas time reached by the latest executed block, as a Unix timestamp.
    • avalanche_evm_transition_sae_gas_time_wall_time_gap_seconds (gauge): gas time minus wall time when the latest block finished executing. A negative value means gas time lags wall time.
    • avalanche_evm_transition_sae_worst_case_base_fee (gauge): worst-case base fee admitted by consensus for the latest executed block.
    • avalanche_evm_transition_sae_executed_base_fee (gauge): base fee realized by execution of the latest executed block.
    • avalanche_evm_transition_sae_worst_case_gas_excess (gauge): worst-case gas excess simulated for the latest executed block.
    • avalanche_evm_transition_sae_executed_gas_excess (gauge): gas excess realized by execution of the latest executed block.
    • avalanche_evm_transition_sae_gas_target (gauge): ACP-176 gas target in force as of the latest executed block.
  • Added avalanche_evm_transition_cchain_min_block_delay_seconds (gauge): ACP-226 minimum block delay currently in force, taken from the most recently executed block.
  • Renamed Coreth and Subnet-EVM state-sync p2p metrics:
    • avalanche_{vmName}_eth_net_tracked_peers -> avalanche_{vmName}_sdk_sync_peer_tracker_num_tracked_peers
    • avalanche_{vmName}_eth_net_responsive_peers -> avalanche_{vmName}_sdk_sync_peer_tracker_num_responsive_peers
    • avalanche_{vmName}_eth_net_average_bandwidth -> avalanche_{vmName}_sdk_sync_peer_tracker_average_bandwidth
  • Added Firewood state-sync proof metrics. All carry the label proof_type="range|change". The duration histograms also carry the label result="success|failure".
    • Server-side histograms: avalanche_{vmName}_sync_server_sync_proof_generation_seconds, avalanche_{vmName}_sync_server_sync_generated_proof_size_bytes, and avalanche_{vmName}_sync_server_sync_proof_shrink_new_key_limit
    • Client-side histograms: avalanche_{vmName}_sync_firewood_sync_proof_verification_seconds, avalanche_{vmName}_sync_firewood_sync_proof_commit_seconds, and avalanche_{vmName}_sync_firewood_sync_received_proof_size_bytes
    • Client-side gauge: avalanche_{vmName}_sync_firewood_sync_request_key_limit

NOTE: {vmName} is evm for Coreth and subnetevm for Subnet-EVM. The sae and cchain metrics exist only on the C-Chain.

Configs

New

  • helicon-min-stake-duration (only on local/custom networks)
  • min-price-target for C-Chain
  • apis for C-Chain. Lists the JSON-RPC APIs that the node serves. See the C-Chain config reference for the names and defaults.
  • api-resolve-pending-to-last-executed for C-Chain. Defaults to true, which resolves pending to the last executed block. Set it to false to resolve pending to the last accepted block.

Changed

  • state-sync-ids in Coreth and Subnet-EVM configs is now a JSON array of node IDs (["NodeID-..."]) instead of a comma-separated string.
  • api-max-duration for C-Chain accepts only a duration string (for example "30s") after Helicon. The node rejects a number. 0 means no limit.
  • state-sync-enabled for C-Chain defaults to true after Helicon.
  • commit-interval for C-Chain must be 4096 on Mainnet and Fuji after Helicon.

Deprecated in C-Chain

  • eth-apis is deprecated. The node maps it onto apis and logs a warning. If the config also sets apis, the node ignores eth-apis. Names whose methods no longer exist, such as admin and debug, log a warning. Any other unknown name is a fatal error. The next release will remove eth-apis.

Removed from C-Chain

After Helicon activates, the C-Chain ignores the following options. The node logs a warning for each unrecognized option and starts.

  • skip-upgrade-check
  • admin-api-enabled
  • admin-api-dir
  • warp-api-enabled
  • continuous-profiler-dir
  • continuous-profiler-frequency
  • continuous-profiler-max-files
  • rpc-gas-cap (fixed at 50,000,000)
  • rpc-tx-fee-cap (fixed at 100 AVAX)
  • trie-dirty-cache
  • trie-dirty-commit-target
  • trie-prefetcher-parallelism
  • preimages-enabled
  • snapshot-wait
  • snapshot-verification-enabled
  • accepted-queue-limit
  • populate-missing-tries-parallelism
  • prune-warp-db-enabled
  • historical-proof-query-window
  • metrics-expensive-enabled
  • price-options-slow-fee-percentage
  • price-options-fast-fee-percentage
  • price-options-max-tip
  • tx-pool-price-limit
  • tx-pool-price-bump
  • tx-pool-lifetime
  • ws-cpu-refill-rate
  • ws-cpu-max-stored
  • allow-unfinalized-queries
  • allow-unprotected-tx-hashes
  • keystore-directory
  • keystore-external-signer
  • keystore-insecure-unlock-allowed
  • push-gossip-percent-stake
  • push-gossip-num-validators
  • push-gossip-num-peers
  • push-regossip-num-validators
  • push-regossip-num-peers
  • push-gossip-frequency
  • pull-gossip-frequency
  • regossip-frequency
  • log-level
  • log-json-format
  • offline-pruning-bloom-filter-size
  • max-outbound-active-requests
  • state-sync-skip-resume
  • state-sync-server-trie-cache
  • state-sync-commit-interval
  • state-sync-min-blocks
  • state-sync-request-size
  • inspect-database
  • accepted-cache-size
  • state-history
  • skip-tx-indexing
  • http-body-limit
  • batch-response-max-size
  • offline-pruning-enabled
  • offline-pruning-data-directory
  • populate-missing-tries
  • transaction-history
  • tx-pool-account-queue
  • tx-pool-global-queue
  • api-max-blocks-per-request

Fixes

  • Updated the minimum Go version from 1.25.8 to 1.25.10.
  • Fixed callTracer failing with incorrect number of top-level calls on transactions where a precompile makes an outbound call (ava-labs/libevm#303).
  • ProposerVM ERROR logs caused by the P-Chain database closing during shutdown are now WARN.
  • Fixed Subnet-EVM nodes failing to restart after a state upgrade activates when the upgrade config contains "storage": {}, "code": "0x", or "balanceChange": "0".
  • Removed the block acceptance time health check that v1.14.2 added. The check could only recover after the node accepted new blocks. Some deployments send traffic only to healthy nodes, so a failed check could never recover.
  • Fixed SizedCache undercounting size when overwriting a key, which let caches exceed their configured limits.

What's Changed

New Contributors

Full Changelog: v1.14.2...v1.15.0

Check out latest releases or
releases around ava-labs/avalanchego v1.15.0

Don't miss a new avalanchego release

NewReleases is sending notifications on new releases.

Get notifications