1.36.4

Notice

This release of nearcore is a security release (CODE_RED_MAINNET). It fixes a handshake parsing vulnerability that, if exploited, can lead to a node crash.

All node operators must upgrade to 1.36.4 immediately.

This release contains no additional code or behavior changes beyond the fix that addresses the security vulnerability, which is related only to a feature that isn't used in production today.

Fixes

• Fix an issue related to handshake parsing

CODE_COLOR: CODE_RED_MAINNET
RELEASE_VERSION: 1.36.4
PROTOCOL_UPGRADE: FALSE
DATABASE_UPGRADE: FALSE
SECURITY_UPGRADE: TRUE

1.36.3

Notice

This release of nearcore is a security release (CODE_RED_MAINNET). It fixes a handshake parsing vulnerability that, if exploited, can lead to a node crash.

All node operators must upgrade to 1.36.3 immediately.

This release contains no additional code or behavior changes beyond the fix that addresses the security vulnerability, which is related only to a feature that isn't used in production today.

Fixes

• Fix an issue related to handshake parsing

CODE_COLOR: CODE_RED_MAINNET
RELEASE_VERSION: 1.36.3
PROTOCOL_UPGRADE: FALSE
DATABASE_UPGRADE: FALSE
SECURITY_UPGRADE: TRUE

1.36.2

Notice

This release of nearcore is a security release (CODE_RED_MAINNET). It fixes a protobuf vulnerability that, if exploited, can lead to a node crash.

All node operators must upgrade to 1.36.2 immediately.

This release contains no additional code or code changes beyond the fix that addresses the security vulnerability.

Fixes

• Fix an issue related to parsing of protobuf messages

CODE_COLOR: CODE_RED_MAINNET
RELEASE_VERSION: 1.36.2
PROTOCOL_UPGRADE: FALSE
DATABASE_UPGRADE: FALSE
SECURITY_UPGRADE: TRUE

1.36.1

Minor CODE_GREEN release, feel free to skip upgrading your nodes.
Adds 2 configuration options and contains no other fixes or improvements.

Fixes

• Add config option tx_routing_height_horizon to configure how many chunk producers are notified about the tx. #10251
• Add config options state_sync.dump.credentials_file to configure pro-active state part production for State Sync. #9487

Crates

Crates corresponding to this nearcore version were published with version 0.18.0

CODE_COLOR: CODE_GREEN_MAINNET
RELEASE_VERSION: 1.36.1
PROTOCOL_UPGRADE: FALSE
DATABASE_UPGRADE: FALSE
SECURITY_UPGRADE: FALSE

1.36.0

CODE_COLOR: CODE_YELLOW_MAINNET
RELEASE_VERSION: 1.36.0
PROTOCOL_UPGRADE: TRUE
DATABASE_UPGRADE: TRUE
SECURITY_UPGRADE: TRUE

Protocol Changes

• The support for fixed shards in shard layout was removed. #9219

Protocol Upgrade Voting

This release contains a protocol upgrade. Voting for upgrading to protocol version 63 will start on 2023-11-13 15:00:00 UTC.

Database Upgrade

This release contains a database upgrade. STATE_SYNC_DUMP_KEY and STATE_SNAPSHOT_KEY was added to the BlockMisc column for support of state sync.

Non-protocol Changes

• New option transaction_pool_size_limit in config.json allows 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

Note around node performance

During internal testing of 1.36 we noticed nodes were impacted by a performance degradation manifesting as slowness in producing blocks. This impact was statistical, and the investigation we performed did not reveal any clear indicator of the root cause of this degradation.
In order to better assess this issue, we decided to test the release on testnet. The testnet release so far did not reveal any performance degradation, so we are releasing to mainnet.
After upgrading on mainnet to 1.36, if you get kicked out or notice any performance issue, please reach out to us on the NEAR Validators Telegram channel with data regarding node configuration and performance.

1.36.0-rc.2

CODE_COLOR: CODE_GREEN_TESTNET
RELEASE_VERSION: 1.36.0-rc.2
PROTOCOL_UPGRADE: FALSE
DATABASE_UPGRADE: FALSE
SECURITY_UPGRADE: FALSE

Protocol Changes

Non-protocol Changes

• Fix an issue that caused unwanted disconnections from peers after 30 minutes #9651, and another that leads to incomplete routing information among peers #9517

1.36.0-rc.1

CODE_COLOR: CODE_YELLOW_TESTNET
RELEASE_VERSION: 1.36.0-rc.1
PROTOCOL_UPGRADE: TRUE
DATABASE_UPGRADE: TRUE
SECURITY_UPGRADE: FALSE

Protocol Changes

• The support for fixed shards in shard layout was removed. #9219

Protocol Upgrade Voting

This release contains a protocol upgrade. Voting for upgrading to protocol version 63 will start on 2023-09-26 15:00:00 UTC.

Database Upgrade

This release contains a database upgrade. STATE_SYNC_DUMP_KEY and STATE_SNAPSHOT_KEY was added to the BlockMisc column for support of state sync.

Non-protocol Changes

• New option transaction_pool_size_limit in config.json allows 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

Note around node performance

During internal testing of 1.36 we noticed nodes were impacted by a performance degradation manifesting as slowness in producing blocks. This impact was statistical, and the investigation we performed did not reveal any clear indicator of the rootcause of this degradation. In order to better assess this issue, we need to determine the behaviour of the 1.36 nodes in a real-life environment, such as testnet. After upgrading on testnet to 1.36, if you get kicked out or notice any performance issue, please reach out to us on the NEAR Validators Telegram channel with data regarding node configuration and performance.

1.35.0

CODE_COLOR: CODE_YELLOW_MAINNET
RELEASE_VERSION: 1.35.0
PROTOCOL_UPGRADE: TRUE
DATABASE_UPGRADE: TRUE
SECURITY_UPGRADE: FALSE

Protocol Changes

• Upgrade the contract preparation code to use finite-wasm, which guarantees deterministic limits on execution time and space of compiled contracts

Non-protocol Changes

• 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_limit config option is set, we now avoid storing more than the specified size of transactions in each shard's transaction pool #8970 and #9036

Split Storage Feature

Split storage allows nodes to improve storage costs and block production performance on archival nodes by splitting storage into hot and cold databases. Hot database stores all of the data from the past several epochs, and cold database stores all historical data from a limited list of columns.
In split storage mode Client only works with a smaller hot database to produce blocks, which results in increased performance.
Cold database is not accessed for reads during normal block production and is only read when historical data is specifically requested. Thus, we recommend keeping the cold database on HDD and only optimize speed for the hot database, which is 10 times smaller.
Split storage can be enabled with a config change and a migration that requires manual steps. You can learn more about the migration process here: https://near-nodes.io/archival/split-storage-archival

Protocol Upgrade Voting

This release contains a protocol upgrade. Voting for upgrading to protocol version 62 will start on 2023-08-01 15:00:00 UTC.

Database Upgrade

This release includes a db upgrade, which migrates entries in the FlatStateChanges, the size of which should not exceed several dozens of entries

1.35.0-rc.1

CODE_COLOR: CODE_YELLOW_TESTNET
RELEASE_VERSION: 1.35.0-rc.1
PROTOCOL_UPGRADE: TRUE
DATABASE_UPGRADE: TRUE
SECURITY_UPGRADE: FALSE

Protocol Changes

• Upgrade the contract preparation code to use finite-wasm, which guarantees deterministic limits on execution time and space of compiled contracts

Non-protocol Changes

• 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_limit config option is set, we now avoid storing more than the specified size of transactions in each shard's transaction pool #8970 and #9036

Split Storage Feature

Split storage allows nodes to improve storage costs and block production performance on archival nodes by splitting storage into hot and cold databases. Hot database stores all of the data from the past several epochs, and cold database stores all historical data from a limited list of columns.
In split storage mode Client only works with a smaller hot database to produce blocks, which results in increased performance.
Cold database is not accessed for reads during normal block production and is only read when historical data is specifically requested. Thus, we recommend keeping the cold database on HDD and only optimize speed for the hot database, which is 10 times smaller.
Split storage can be enabled with a config change and a migration that requires manual steps. We have several choices for the migration:
https://docs.google.com/document/d/1PVuETy4Ogf_8Hl0nFRrIvBs-94OhSC1mrULjsQcH-Bc/edit?usp=sharing

Protocol Upgrade Voting

This release contains a protocol upgrade. Voting for upgrading to protocol version 62 will start on 2023-06-27 15:00:00 UTC.

Database Upgrade

This release includes a db upgrade, which migrates entries in the FlatStateChanges, the size of which should not exceed several dozens of entries

1.34.0

CODE_COLOR: CODE_YELLOW_MAINNET
RELEASE_VERSION: 1.34.0
PROTOCOL_UPGRADE: TRUE
DATABASE_UPGRADE: TRUE
SECURITY_UPGRADE: TRUE

Protocol Changes

• Flat Storage for reads. Reduces the number of DB accesses for state reads from 2 * key.len() in the worst case to 2. #8761, NEP-399
• 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

Flat Storage Migration

This is a very important migration that needs special attention.
The migration starts automatically after switching to the 1.34.0 binary and runs in the background.
The migration completes approximately in:

• RPC nodes: 11-15 hours
• Archival nodes: 30-40 hours

The migration increases disk usage by approximately 10 GB.
Please upgrade early to make sure there is enough time for the migration to complete.
Note that the protocol voting date is set further into the future than usually, to make sure that all RPC and Archival nodes have an opportunity to upgrade.

Failure to finish the migration before the protocol upgrade will result in the node getting stuck.
The migration cannot be started if the node has already missed the protocol upgrade.

To check that the migration has completed, check the metrics page, e.g. http://127.0.0.1:3030/metrics .
If flat_storage_creation_status has value of 2 for each of the shards, then the migration is complete.
If flat_storage_creation_status is missing and flat_storage_distance_to_head has a very low value for each of the shards, then the migration is complete.

Reduce Memory Usage

If you are running a node on minimal hardware specifications, you can save 2GB of RAM by decreasing store.trie_cache.per_shard_max_bytes.s3.v1 from 3000000000 (default value) to 1000000000.
All our testing shows that things work stable with that and block processing stays below 1 second.
However, keeping store.trie_cache.per_shard_max_bytes.s3.v1=3000000000 has a positive impact on reducing the processing time and therefore gives an extra safety buffer.

Protocol Upgrade Voting

Voting for upgrading to protocol version 61 will start on 2023-06-12 15:00:00 UTC. Please upgrade early because of a special Flat Storage migration.

Database Upgrade

This release includes a trivial DB upgrade. It changes no data, but informs about the Flat Storage migration. The upgrade completes instantly and requires no extra disk space.
Several new columns will be created by the Flat Storage migration in the background.

Non-Protocol Changes

• 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

Fixes

• Extra validation of DeleteAccountAction