- The contract runtime has been upgraded to use the new Wasmtime-based runtime;
- The contract runtime now allows for bulk memory instructions in Wasm code.
No Changes
- Introducing deterministic account IDs: Account IDs starting with the prefix
0scan only be created with the new actionDeterministicStateInit. Its name is derived from the initial state and global contract it uses. #14307 - New host functions for deterministic initialization: Three new host functions
promise_batch_action_state_init,promise_batch_action_state_init_by_account_idandset_state_init_data_entryallow creating the new actionDeterministicStateInitfrom within a smart contract. #14364 - Accessing the current contract code: The new host function
current_contract_codeallows smart contracts to read the currently deployed code hash or global contract identifier. #14372 - Controlling balance refunds: The new host function
promise_set_refund_toallows smart contracts to redirect balance refunds of outgoing receipts to other accounts. #14285 - Querying refund receivers: The new host function
refund_to_account_idreturns the receiver of balance refunds, which is eitherpredecessor_idor the refund receiver set by the predecessor usingpromise_set_refund_to. #14372 - Gas optimization: Calls to the existing host functions
inputandpromise_resultno longer charge gas per byte (wasm_write_memory_byte), thanks to an optimization that eliminates unnecessary data copying. #14405
- Indexer changes, including breaking changes in the API. See the indexer changelog for details.
No Changes
- Moved Tier1 configuration from experimental to top level config. No action is necessary as the default values are the recommended ones. (#13575)
- Add a new configuration option
save_tx_outcomes(#13610). When set tofalse, per-transaction outcomes are not written to the db to improve validator throughput. Disabling this config means transactions processed by the node will not be queryable by transaction hash, however this is not needed for validators to perform their duties. The default for archive and RPC nodes istrue. - Updated the recommended operating system network settings for running
neard(#14012).
- Increase number of validators from 300 to 500. To maintain the chain safety, we increase the number of mandates per shard to 105. (#14033)
- Added a neard subcommand tool that can be used to recover data that was lost due to bug in resharding. (#14185)
- Add a new configuration option
protocol_version_check_config_override. This determines which epoch the client uses to determine version compatibility with the network. The default isNextNext, which causes the node to exit instead of persisting potentially incorrect epoch info and corrupting the database in case the node operator does not update the client in time for the network upgrade. The prior behavior only checked the protocol version in the next epoch, which can be restored by setting this option toNextand allows a validator that did not upgrade their node to participate in the final epoch the client is compatible with the network.
- A new shard layout for production networks (#13324). Use split boundary from #13609.
- When the protocol update version voting takes place, validators that did not upgrade to the latest version will be scheduled for removal (aka kickout) in the epoch the new version takes effect. This helps avoid missed blocks in the first epoch of the new version, as un-upgraded validators would produce invalid blocks. Technically this is a protocol change as it impacts the validator set, however it will take effect during the next version upgrade therefore does not require its own protocol version. (#13375)
- Implement NEP-536: Reduce the number of refund receipts by adding removing pessimistic gas pricing. Also introduce a gas refund penalty but set it to 0 to avoid potential negative impact. (#13397)
- Implement P2P sync for state sync headers. (#13377)
- Enable saturating float-to-int conversions in runtime. (#13414)
- Add RPC query for viewing global contract code. (#13547)
- Add promise batch host functions for global contracts. (#13565)
- Stabilize
EXPERIMENTAL_changesRPC method and rename it tochanges. (#13722) - Rename
TxRequestHandlerActortoRpcHandlerActorto reflect the change in the scope of its responsibilities. Otherwise its API change is fully backward-compatible, so the dependent services can handle it by simply renaming the type where it is mentioned explicitly. (#13259)
- Implemented support for global contracts: NEP-591
- Implemented Optimistic Block to remove doubled chunk execution latency from block & chunk production flow #10584
- Changed receipt id computation to enable chunk execution based on Optimistic Block. More specifically, primitive
create_hash_upgradableis changed to useblock_heightinstead ofextra_hash.
No Changes
- Add cross-shard bandwidth scheduler which manages transferring receipts between shards, enabling higher throughput of cross-shard receipts and better horizontal scalability. NEP-584 (near/NEPs#584)
- Resharding V3 - a new implementation for resharding and two new shard layouts for the production networks. NEP-568 (near/NEPs#568)
- Parallelize transaction validation (including signature checks) before
verify_and_charge_transaction, significantly improving throughput for transaction processing on the nodes. #12654 - Current Epoch State Sync - Moves the sync point from the previous epoch to the current epoch. #12102
- Fixing invalid cost used for
wasm_yield_resume_byte. #12192 - Relaxing Congestion Control to allow accepting and buffering more transactions. #12241 #12430
- Exclude contract code out of state witness and distribute it separately. #11099
- Epoch Sync V4: A capability to bootstrap a node from another active node. #73
- Decentralized state sync: Before, nodes that needed to download state (either because they're several epochs behind the chain or because they're going to start producing chunks for a shard they don't currently track) would download them from a centralized GCS bucket. Now, nodes will attempt to download pieces of the state from peers in the network, and only fallback to downloading from GCS if that fails. Please note that in order to participate in providing state parts to peers, your node may generate snapshots of the state. These snapshots should not take too much space, since they're hard links to database files that get cleaned up on every epoch. #12004
- Sets
chunk_validator_only_kickout_thresholdto 70. Uses this kickout threshold as a cutoff threshold for contribution of endorsement ratio in rewards calculation: if endorsement ratio is above 70%, the contribution of endorsement ratio in average uptime calculation is 100%, otherwise it is 0%. Endorsements received are now included inBlockHeaderto improve kickout and reward calculation for chunk validators.
- Added documentation and a reference script to recover the data lost in archival nodes at the beginning of 2024.
- Archival nodes only: Stop saving partial chunks to
PartialChunkscolumn in the Cold DB. Instead, archival nodes will reconstruct partial chunks from theChunkscolumn. - Enabled state snapshots on every epoch to allow the nodes to take part in decentralized state sync in future releases.
This release patches a bug found in the 2.2.0 release
There was a bug in the integration between ethereum implicit accounts and the compiled contract cache which sometimes caused the nodes to get stuck. This would most often happen during state sync, but could also happen by itself. Please update your nodes to avoid getting stuck.
A node that hits this bug will print an error about an InvalidStateRoot in the logs and then it'll be unable to sync.
It's possible to recover a stalled node by clearing the compiled contract cache and rolling back one block:
- Stop the neard process
- Download the new version of neard
- Clear the compiled contract cache: rm -rf ~/.near/data/contracts
- Undo the last block: ./neard undo-block
- Start neard
After that the node should be able to recover and sync with the rest of the network.
- The minimum validator stake has been set to a lower value. The small-stake validators that were kicked out during the shift to stateless validation will be able to rejoin the network.
- Better algorithm for validator kickouts
- (Testnet only) update the eth-implicit accounts contract on testnet to match the one on mainnet.
- Fix spammy messages about calculating gas for PromiseYield receipts.
- Don't crash when the CPU doesn't have SHA-NI instructions. It's still a hardware requirement, there is no guarantee that nodes without this instruction will be able to keep up with the network, but
neardwill now be able to run (slowly) on CPUs without this instruction.
-
Enforce rate limits to received network messages #11617. Rate limits are configured by default, but they can be overridden through the experimental configuration option
received_messages_rate_limits. -
Increase sync blocks requested and make it configurable through a parameter in the client config #11820. Increase default max sync block requests to 10 from 5.
No Changes
No Changes
- Use more precise gas costs for function calls #10943 that should lead to more efficient chunk utilization.
- Limit overcharging by decoupling minimum_new_receipt_gas from the function call and setting it to a constant value. #10941
- These PRs introduce a change in the default behaviour of
broadcast_tx_commit,send_tx,tx, EXPERIMENTAL_tx_statusRPC methods. The default behaviour no longer waits for refund receipts. If you do need to wait for refund receipts, you need ask about it explicitly by using TxExecutionStatus::Final option ("wait_until": "FINAL" in the json request). More information in the #10792 #10948 - Adds improvements to the
sweatcontract prefetcher logic. Add new prefetcher logic forkaichingcontract. #10899 - Improves prefetcher logic to speedup chunk finalization by prefetching keys related to refund receipts and actions such as: Delegate, AddKey, DeleteKey. #10936
- Add more metrics for receipt processing. #10944
- Resharding v2 - new implementation for resharding and a new shard layout for production networks. #10303, NEP-0508
- Restrict the creation of non-implicit top-level account that are longer than 32 bytes. Only the registrar account can create them. #9589
- Adjust the number of block producers and chunk producers on testnet to facilitate testing of chunk-only producers #9563
- Add prometheus metrics for the internal state of the doomslug. #9458
- Fix
EXPERIMENTAL_protocol_configto apply overrides fromEpochConfig. #9692 - Add config option
tx_routing_height_horizonto configure how many chunk producers are notified about the tx. #10251
- The support for fixed shards in shard layout was removed. #9219
- New option
transaction_pool_size_limitinconfig.jsonallows to limit the size of the node's transaction pool. By default the limit is set to 100 MB. #3284 - Database snapshots at the end of an epoch. This lets a node obtain state parts using flat storage. #9090
- Number of transactions included in a chunk will be lowered if there is a congestion of more than 20000 delayed receipts in a shard. #9222
- Our more efficient and scalable V2 routing protocol is implemented. It shadows the V1 protocol for now while we verify its performance. #9187
- The default config now enables TIER1 outbound connections by default. #9349
- State Sync from GCS is available for experimental use. #9398
- Upgrade the contract preparation code to use finite-wasm, which guarantees deterministic limits on execution time and space of compiled contracts
- Dump state by multiple nodes, each node will refer to s3 for which parts need to be dumped. #9049
- Small values in the flat storage trie are inlined for faster accesses #9029
- A current protocol version metric is added to the prometheus metrics under near_current_protocol_version #9030
- The transaction pool size is tracked, and if the
transaction_pool_size_limitconfig option is set, we now avoid storing more than the specified size of transactions in each shard's transaction pool #8970 and #9036
- Flat Storage for reads, reducing number of DB accesses for state read from
2 * key.len()in the worst case to 2. #8761, NEP-399 - Contract preparation and gas charging for wasm execution also switched to using our own code, as per the finite-wasm specification. Contract execution gas costs will change slightly for expected use cases. This opens up opportunities for further changing the execution gas costs (eg. with different costs per opcode) to lower contract execution cost long-term. #8912
- Compute Costs are implemented and stabilized. Compute usage of the chunk is now limited according to the compute costs. #8915, NEP-455.
- Write related storage compute costs are increased which means they fill a chunk sooner but gas costs are unaffected. #8924
- undo-block tool to reset the chain head from current head to its prev block. Use the tool by running:
./target/release/neard undo-block. #8681 - Add prometheus metrics for expected number of blocks/chunks at the end of the epoch. #8759
- Node can sync State from S3. #8789
- Node can sync State from local filesystem. #8913
- Add per shard granularity for chunks in validator info metric. #8934
- State-viewer tool to dump and apply state changes from/to a range of blocks. #8628
- Experimental option to dump state of every epoch to external storage. #8661
- Add prometheus metrics for tracked shards, block height within epoch, if is block/chunk producer. #8728
- State sync is disabled by default #8730
- Node can restart if State Sync gets interrupted. #8732
- Merged two
neard view-statecommands:apply-state-partsanddump-state-partsinto a singlestate-partscommand. #8739 - Add config.network.experimental.network_config_overrides to the JSON config. #8871
- Fix: rosetta zero balance accounts #8833
- Fix vulnerabilities in block outcome root validation and total supply validation #8790
- Stabilize
ed25519_verifyfeature: introducing a host function to verify ed25519 signatures efficiently. #8098 NEP-364 - Added STUN-based self-discovery to make configuration of TIER1 network easier in the simplest validator setups. #8472
- Stabilize zero balance account feature: allows account to not hold balance under certain conditions and enables a more smooth onboarding experience where users don't have to first acquire NEAR tokens to pay for the storage of their accounts. #8378 NEP-448
- Stabilize meta transactions on the protocol level. NEP-366, Tracking issue #8075, Stabilization #8601
- Config validation can be done by following command:
./target/debug/neard --home {path_to_config_files} validate-config. This will show error if there are file issues or semantic issues inconfig.json,genesis.json,records.json,node_key.jsonandvalidator_key.json. #8485 - Comments are allowed in configs. This includes
config.json,genesis.json,node_key.jsonandvalidator_key.json. You can use//,#and/*...*/for comments. #8423 /debugpage now has client_config linked. You can also check your client_config directly at /debug/client_config #8400- Added cold store loop - a background thread that copies data from hot to cold storage and a new json rpc endpoint - split_storage_info - that exposes debug info about the split storage. #8432
ClientConfigcan be updated while the node is running.dyn_config.jsonis no longer needed as its contents were merged intoconfig.json. #8240- TIER2 network stabilization. Long-lasting active connections are persisted to DB and are re-established automatically if either node restarts. A new neard flag
--connect-to-reliable-peers-on-startupis provided to toggle this behavior; it defaults to true. The PeerStore is no longer persisted to DB and is now kept in-memory. #8579, #8580.
- Enable TIER1 network. Participants of the BFT consensus (block & chunk producers) now can establish direct TIER1 connections between each other, which will optimize the communication latency and minimize the number of dropped chunks. To configure this feature, see advanced_configuration/networking. #8141 #8085 #7759
- [Network] Started creating connections with larger nonces, that are periodically refreshed Start creating connections (edges) with large nonces #7966
/statusresponse has now two more fields:node_public_keyandvalidator_public_key. Thenode_keyfield is now deprecated and should not be used since it confusingly holds validator key. #7828- Added
near_node_protocol_upgrade_voting_startPrometheus metric whose value is timestamp when voting for the next protocol version starts. #7877 - neard cmd can now verify proofs from JSON files. #7840
- In storage configuration, the value
trie_cache_capacitiesnow is no longer a hard limit but instead sets a memory consumption limit. For large trie nodes, the limits are close to equivalent. For small values, there can now fit more in the cache than previously. #7749 - New options
store.trie_cacheandstore.view_trie_cacheinconfig.jsonto set limits on the trie cache. Deprecates the never announcedstore.trie_cache_capacitiesoption which was mentioned in previous change. #7578 - New option
store.background_migration_threadsinconfig.json. Defines number of threads to execute background migrations of storage. Currently used for flat storage migration. Set to 8 by default, can be reduced if it slows down block processing too much or increased if you want to speed up migration. #8088, - Tracing of work across actix workers within a process: #7866, #7819, #7773.
- Scope of collected tracing information can be configured at run-time: #7701.
- Attach node's
chain_id,node_id, andaccount_idvalues to tracing information: #7711. - Change exporter of tracing information from
opentelemetry-jaegertoopentelemetry-otlp: #7563. - Tracing of requests across processes: #8004.
- Gas profiles as displayed in the
EXPERIMENTAL_tx_statusare now more detailed and give the gas cost per parameter.
- Stabilize
account_id_in_function_call_permissionfeature: enforcing validity of account ids in function call permission. #7569
use_db_migration_snapshotanddb_migration_snapshot_pathoptions are now deprecated. If they are set inconfig.jsonthe node will fail if migration needs to be performed. Usestore.migration_snapshotinstead to configure the behaviour #7486- Added
near_peer_message_sent_by_type_bytesandnear_peer_message_sent_by_type_totalPrometheus metrics measuring size and number of messages sent to peers. #7523 near_peer_message_received_totalPrometheus metric is now deprecated. Instead of it aggregatenear_peer_message_received_by_type_totalmetric. For example, to get total rate of received messages usesum(rate(near_peer_message_received_by_type_total{...}[5m])). #7548- Few changes to
view_stateJSON RPC query:- The request has now an optional
include_proofargument. When set totrue, response’sproofwill be populated. - The
proofwithin each value invalueslist of aview_stateresponse is now deprecated and will be removed in the future. Client code should ignore the field. - The
prooffield directly withinview_stateresponse is currently always sent even if proof has not been requested. In the future the field will be skipped in those cases. Clients should accept responses with this field missing (unless they setinclude_proof). #7603
- The request has now an optional
- Backtraces on panics are enabled by default, so you no longer need to set
RUST_BACKTRACE=1environmental variable. To disable backtraces, setRUST_BACKTRACE=0. #7562 - Enable receipt prefetching by default. This feature makes receipt processing
faster by parallelizing IO requests, which has been introduced in
#7590 and enabled by default
with #7661.
Configurable in
config.jsonusingstore.enable_receipt_prefetching.
- Stabilized
protocol_feature_chunk_only_producers. Validators will now be assigned to blocks and chunks separately. - The validator uptime kickout threshold has been reduced to 80%
- Edge nonces between peers can now optionally indicate an expiration time
- The logic around forwarding chunks to validators is improved
- Approvals and partial encoded chunks are now sent multiple times, which should reduce stalls due to lost approvals when the network is under high load
- We now keep a list of "TIER1" accounts (validators) for whom latency/reliability of messages routed through the network is critical
- /debug HTTP page has been improved
- Messages aren't routed through peers that are too far behind
- Log lines printed every 10 seconds are now less expensive to compute
- message broadcasts have been improved/optimized
network.external_addressfield in config.json file is deprecated. In fact it has never been used and only served to confuse everyone #7300- Due to increasing state size, improved shard cache for Trie nodes to put more nodes in memory. Requires 3 GB more RAM #7429
- Stabilized
alt_bn128_g1_multiexp,alt_bn128_g1_sum,alt_bn128_pairing_checkhost functions #6813.
- Added
pathoption toStoreConfigwhich makes location to the RocksDB configurable viaconfig.jsonfile (atstore.pathpath) rather than being hard-coded todatadirectory in neard home directory #6938 - Removed
testnetalias forlocalnetcommand; it’s been deprecated since 1.24 #7033 - Removed undocumented
unsafe_reset_allandunsafe_reset_datacommands; they were deprecated since 1.25 - Key files can use
private_keyfield instead ofsecret_keynow; this improves interoperability with near cli which uses the former name #7030 - Latency of network messages is now measured #7050
- Introduced protobuf encoding as the new network protocol. Borsh support will be removed in two releases as per normal protocol upgrade policies #6672
- Added
near_peer_message_received_by_type_bytes#6661 andnear_dropped_message_by_type_and_reason_count#6678 metrics. - Removed
near_<msg-type>_{total,bytes}#6661,near_<msg-type>_dropped,near_drop_message_unknown_accountandnear_dropped_messages_count#6678 metrics. - Added
near_action_called_countmetric [#6679]((near#6679) - Removed
near_action_<action-type>_totalmetrics [#6679]((near#6679) - Added
near_build_infometric which exports neard’s build information #6680 - Make it possible to update logging at runtime: #6665
- Use correct cost in gas profile for adding function call key #6749
- Enable access key nonce range for implicit accounts to prevent tx hash collisions #5482
- Include
promise_batch_action_function_call_weighthost function on the runtime #6285 #6536 - Increase deployment cost #6397
- Limit the number of locals per contract to 1_000_000
- Ensure caching all nodes in the chunk for which touching trie node cost was charged, reduce cost of future reads in a chunk #6628
- Lower storage key limit to 2 KiB
- Switch to LZ4+ZSTD compression from Snappy in RocksDB #6365
- Moved Client Actor to separate thread - should improve performance #6333
- Safe DB migrations using RocksDB checkpoints #6282
- NEP205: Configurable start of protocol upgrade voting #6309
- Make max_open_files and col_state_cache_size parameters configurable #6584
- Make RocksDB block_size configurable #6631
- Increase default max_open_files RocksDB parameter from 512 to 10k #6607
- Use kebab-case names for neard subcommands to make them consistent with flag names. snake_case names are still valid for existing subcommands but kebab-case will be used for new commands.
max_gas_burnthas been increased to 300.
- More Prometheus metrics related to epoch, sync state, node version, chunk fullness and missing chunks have been added.
- Progress bar is now displayed when downloading
config.jsonandgenesis.json. - Status line printed in logs by
neardis now more descriptive.
view_stateis now a command ofneard;state-vieweris no longer a separate binary.RUST_LOGenvironment variable is now correctly respected.NetworkConfig::verifywill now fail if configuration is invalid rather than printing error and continuing.- Fixed a minor bug which resulted in DB Not Found errors when requesting chunks.
- Updated to wasmer-near 2.2.0 which fixes a potential crash and improves cost estimator working.
neard initwill no longer override node or validator private keys.- Rosetta RPC now populates
related_transactionsfield. - Rosetta RPC support is now compiled in by default. The feature still needs to be explicitly turned on and is experimental.
- Rosetta RPC /network/status end point correctly works on non-archival nodes.
unsafe_reset_allandunsafe_reset_datacommands are now deprecated. Usermexplicitly instead.
- Enable access key nonce range for implicit accounts to prevent tx hash collisions.
- Upgraded our version of pwasm-utils to 0.18 -- the old one severely under-counted stack usage in some cases.
- Fix a bug in chunk requesting where validator might request chunks even if parent block hasn’t been processed yet.
- Fix memory leak in near-network.
- Change block sync to request 5 blocks at a time
- Change NUM_ORPHAN_ANCESTORS_CHECK to 3
- Further lower regular_op_cost from 2_207_874 to 822_756.
- Limit number of wasm functions in one contract to 10_000. #4954
- Add block header v3, required by new validator selection algorithm
- Move to new validator selection and sampling algorithm. Now we would be able to use all available seats. First step to enable chunk only producers.
- Increase RocksDB cache size to 512 MB for state column to speed up blocks processing #5212
- Upgrade from Wasmer 0 to Wasmer 2, bringing better performance and reliability. #4934
- Lower regular_op_cost (execution of a single WASM instruction) from 3_856_371 to 2_207_874. #4979
- Lower data receipt cost and base cost of
ecrecoverhost function. - Upgrade from one shard to four shards (Simple Nightshade Phase 0)
- Fix some receipts that were stuck previously due to #4228. #4248
- Improve contract module serialization/deserialization speed by 30% #4448
- Make
AccountIdstrictly typed and correct by construction #4621 - Address test dependency issue #4556 #4606. #4622.
- Fix neard shutdown issue #4429. #4442
- Introduce new host functions
ecrecoverandripemd160. #4380 - Make
Accounta versioned struct. #4089 - Limit the size of transactions to 4MB. #4107
- Cap maximum gas price to 20x of minimum gas price. #4308, #4382
- Fix
storageUsagefor accounts that were affected by #3824. #4272 - Fix a bug in computation of gas for refunds. #4405