Spring v1.1.0-rc2

The second release candidate for Spring v1.1.0 fixes an issue discovered during replay testing.

Note

Any node that had previously upgraded to v1.1.0-rc1 should restart from a v1.0.x snapshot after upgrading to v1.1.0-rc2.

Complete Change Log

• [1.1.0] Fix deferred trx processing by @heifner in #1150
• [1.1.0] Do not allow replay-blockchain without snapshot by @heifner in #1156
• [1.1.0] Fix account billing on eos-vm-oc interrupt by @heifner in #1160
• [1.1.0] Update to 1.1.0-rc2 by @heifner in #1162

Full Changelog: v1.1.0-rc1...v1.1.0-rc2

Spring v1.1.0-rc1

Warning

v1.1.0-rc1 has a known issue that impacts the integrity of state. Do not upgrade to rc1 (wait for next release).

Spring v1.1.0-rc1 reduces bandwidth usage, streamlines block handling, and adds support for designated smart contracts to leverage EOS VM OC for higher performance.

Notable Changes

• Optimized P2P Block Propagation: Introduced a Block Nack feature to reduce bandwidth usage and enhance sync efficiency during block propagation.
• Enhanced Block Processing Efficiency: several improvements, including deduplication of redundant posts and notifications for new blocks.
• Pause at Block: Introduced the ability to pause execution at a specified block height. This allows operators to halt processing at a predefined block for debugging, upgrades, or controlled synchronization.
• Optimized EOS VM OC for Designated Smart Contracts: Added an option to whitelist specific account suffixes for EOS VM OC execution during block production.

Upgrade Procedure

Upgrading from Spring 1.0.1+ requires no manual upgrade procedure. Simply stop nodeos, install 1.1.0, and relaunch nodeos.

New Config Options

Manage OC Whitelist

eos-vm-oc-whitelist

• Description: Specifies account suffixes to whitelist for eos-vm-oc, allowing designated smart contracts to bypass subjective limits and improve execution performance.
• Usage:
  The default value of eos-vm-oc-whitelist is xsat. If xsat is not desired, then an alternative must be specified. If there is a desire to not whitelist any accounts, then use eosio.null or any other account without a contract. Restart your node for changes to take effect.
• Example:
  To whitelist accounts ending in .suffix1 and .suffix2, and remove the default value of ".xsat" accounts:

  eos-vm-oc-whitelist suffix1 suffix2

Disable Optimized Block Propagation

p2p-disable-block-nack

• Description: The Block Nack feature optimizes P2P communication by reducing bandwidth usage and improving block synchronization efficiency.
• Usage:
  The Block Nack feature is enabled by default. It can be disabled by adding the p2p-disable-block-nack = true option to your configuration file (config.ini) and restart your node for changes to take effect.
• Example:
  To disable efficient block propagation:

  p2p-disable-block-nack = true

Specify a Transactions-Only Endpoint

p2p-listen-endpoint

• Description: Configures a P2P endpoint to accept only transaction traffic.
• Configuration:
  Add the following to your config.ini or command line:

  p2p-listen-endpoint = <host>:<port>:trx
  

• Example:
  To create a transactions-only endpoint on port 9876:

  p2p-listen-endpoint = 0.0.0.0:9876:trx
  

Complete Change Log

Optimized Block Propagation

• Optimize P2P Block Propagation with Block Nack Feature: Reduces Bandwidth Overload and Improves Sync Efficiency
• Simplify and improve duplicate connection logic.
• Update net-plugin to prioritize blocks, votes, and messages to ensure timeline arrival of critical information.
• Add test net to performance harness.
• Use net threads for connecting and timers.
• With the exception of unlinkable blocks, disconnect immediately instead of allowing for 13 invalid blocks before disconnect.
• Improve P2P performance by maintaining serialized format from block log to peer syncing.
• P2P: Add p2p-listen-endpoint :trx & :blk support
• P2P: Do not log expected conditions in send callback

Optimized Block Processing

• Optimize block processing - Part 1.
• Optimize block processing - Part 2: notifying only when new blocks are ready for processing.
• Optimize block processing - Part 3: deduplication of redundant posts.
• Enhance robustness of speculative block mode by allowing for future blocks. Future blocks commonly from clock-skew.
• Test for producer_plugin pause_at_block

OC for Designated Smart Contracts

• Support interrupt transaction, brings along app base submodule update.
• Remove subjective eos-vm-oc limits for whitelisted accounts.
• Add option config eos-vm-oc-whitelist which allows for specification of whitelisted account suffixes for eos-vm-oc.
• Always interrupt and restart long-running transactions once OC compile completes.
• Interrupt block validation on new best head, enables auto recovery from poison blocks.
• Make eos-vm-oc-whitelist option multitoken
• workaround cdt's small const memcpy host function calls in EOS VM OC
• Allow the dev&test synchronous EOS VM OC mode without adjusting a compile time option
• Don't pause billing timer when fetching EOS VM OC's code_descriptor

Logging

• Correct ilog messages for pre-Savanna produced block.
• Clean up redundant log and metrics output during replay.
• Add details to wasm_execution_error message..
• On startup log number of blocks to be applied from the fork db.
• Add the ability for deepmind to go to stderr, which is unbuffered.

Tests

• Added a suite of tests to verify SHiP recovers when restarted in a variety of log and index corruption scenarios.
• Eliminate spurious test failures by relaxing contract cache timing
• Retain logs when plugin tests fail.
• Resolve test instability in distributed transaction tests.
• Add test for invalid QC block number.
• Modify python integration test Node.waitForBlock to wait on blockNum instead of blockNum+1.
• Fix test contract EOSIO_COMPILE_TEST_CONTRACTS, by removing unused EOSIO_DISPATCH
• Add unit test for find_secondary host functions.
• Regenerate jumborow.wasm when EOSIO_COMPILE_TEST_CONTRACTS enabled.
• Capture test output for debugging purposes.
• Refactor checktime and separate out those tests.
• Cleanup tests, removed tests for Leap 3.1 and below. Removed tests for history_api_plugin which does not exist in Spring.
• Clean up Performance Harness tests, removed logic for versions of Leap 2.x.
• Update string for assert text.
• Update ordered_diff tests to check the diff result is not changing.
• Test additional versions 7,8, and 9 of LLVM.
• Relax eos vm test limits on generated code size.
• Test more versions of LLVM.
• fix Python 3.12 SyntaxWarning due to invalided escape sequence.
• Add tests for get_serialized_block_by_number.
• Resolve unused warning in libtester.
• Improve div and mod operations in webassemble to reduce overflows.
• Add compiler buildtins tests to Spring. Previously included in Leap CI.
• Adds finality transition tests modeled after the fastnet transition tests.
• Test: Create unique log files for nodeos
• Test: Increase shutdown timeout
• [Performance Harness: Change default transaction generator trx expiration](https://github.com/A...

Spring v1.0.4

Spring v1.0.4 is a patch release focused on system stability, performance enhancements, and accurate logging to facilitate better monitoring and maintenance.

Note

This release patches a security vulnerability affecting API nodes. All public API nodes should upgrade to mitigate a possible denial of service vector.

Notable Changes

• Fast Shutdown in Irreversible Mode: Implemented a mechanism for swift and safe system termination when operating in irreversible mode.
• Memory Usage During Sync: Advance fork database root correctly when the LIB exceeds HEAD during sync, keeping memory within expected limits.
• Smoother Syncing with Tight Fetch Span: Resolved issues causing nodes to become stuck during synchronization with tight sync-fetch-span settings.
• Accurate Peer Connection Logging: Corrected the peer connections count in logs to reflect only active connections.
• Security Patch: public API nodes should upgrade to avoid a possible denial of service vector.

Complete Change Log

Performance

• Fast shutdown in irreversible mode.
• Fix fetch block with no block log. This removes unneeded warning by fixing underlying issue.

P2P

• Fix sync stuck that can occur with very tight sync-fetch-span.
• Fix stuck plugin http test.
• More accurate count of peer connections.

Tests

• Allow for block to not have qc_extension
• Fix nodeos_startup_catchup tests.
• Optimize tests by skipping EOS VM interpreter unit tests on sanitizer builds.

Other

• Advance fork db root when pending lib greater than head.
• Fix spurious access violations and bump eos-vm submodule.
• Set CDT to latest release version

PR List

Contributors

Special thanks to the contributors that submitted patches for this release:

• @spoonincode
• @heifner
• @linh2931

Spring v1.0.3

Spring v1.0.3 is a patch release focusing on stability, performance, and compatibility improvements. It brings key fixes to the Trace API, enhancements to P2P networking, optimized block processing during syncing, and expanded platform support.

Notable Changes

Trace API Fix: Corrected the get_transaction_trace endpoint to properly handle forked blocks.

P2P Stability Improvements: Improved reconnection handling, fixed syncing timeout issue, and optimized processing of invalid vote messages for better network stability.

Optimized block processing: Enhanced responsiveness during syncing—allows quicker shutdowns, faster get_info API responses, and restores periodic logging during replay.

Compatibility Improvements: Resolved build issues with LLVM versions 7–9 and fixed some unit tests on non-x86 and non-Linux platforms.

Complete Change Log

Bug Fixes:

• Fixed P2P stability issue caused by closing file descriptor 1 instead of connection when syncing timed out. (#897)
• Fixed get_transaction_trace endpoint: Addressed an issue in the Trace API where the endpoint failed to return transaction traces if the initial block forked out, ensuring accurate trace retrieval. (#966)

P2P Networking Enhancements:

• Improved reconnection handling: Resolved issues with P2P reconnections to enhance network stability and reliability. (#879, #899)
• Optimized invalid vote message handling: Streamlined processing of invalid vote messages to reduce unnecessary load and improve network efficiency. (#913)

Performance Improvements:

• Optimized block application: Implemented limits on block application to improve performance during block processing. (#921)

Logging Improvements:

• Updated Prometheus metrics: Enhanced speculative block metrics for produced blocks, providing more accurate monitoring data. (#957)
• Enhanced logging accuracy: Corrected block report log times to exclude the duration of maybe_switch_forks, ensuring precise logging information. (#881)
• Cleaned up logging during restarts: Fixed an issue where restarting with Ctrl-C during fork database replay caused excessive log spamming, ensuring cleaner logs and smoother restarts. (#894)

Testing Improvements:

• Improved test reliability: Updated flaky HTTP plugin unit tests and fixed the production_pause_max_rev_blks_test, strengthening the test suite's stability. (#959, #960)
• Resolved flaky tests due to race conditions: Fixed race conditions causing unreliable tests like nodeos_snapshot_diff_test, p2p_sync_throttle_test, and distributed-transactions-if-test, improving test reliability. (#875, #897, #908)
• Cross-platform test support: Updated eosvmoc_limits_tests to function on non-x86 and non-Linux platforms where OC is not supported, enhancing cross-platform compatibility. (#904)

Other Changes:

• Enhanced compiler compatibility: Fixed building issues with LLVM versions 7, 8, and 9, expanding support for different compiler environments. (#903)
• Version tracking update: Merged version bump from 5.0.3 to 1.0.3 for consistent tracking across releases. (#877)
• Fork handling improvements: Reviewed and implemented alternative methods for handling forks to improve fork management. (#968)

Contributors

Special thanks to the contributors that submitted patches for this release:

• @spoonincode
• @greg7mdp
• @linh2931
• @heifner

Spring v1.0.2

This release focuses on enhancing stability and performance, with improved block synchronization, logging, and system compatibility.

Notable Changes

• Support for Linux Kernel 4.15: Added compatibility for systems running Linux Kernel 4.15 by falling back to random_access_file, ensuring broader platform support. (#780)
• P2P synchronization enhancements: P2P connections continue during snapshot operations, improved reconnection handling, prevented syncing ahead in irreversible mode, and continue syncing when last irreversible block (LIB) is paused. (#768, #838, #811, #777)
• State History ABI updates: Added block header extension types to the state history ABI, enhancing developers' ability to work with extended block data. (#783)
• Improved logging accuracy: Enhanced the reporting of produced and received block times, disambiguated time entries in logs, and updated handshake logs to debug level to reduce verbosity. (#850, #837, #839)
• Fixed hangs on shutdown and exit: Resolved issues that caused the node to hang during shutdown processes, ensuring smoother and more reliable operations. (#845, #816)

Complete Change Log

SHiP

• Fix hang on exit, draining references to SHiP found in io_service.
• Fix hang on shutdown.
• Add block header extensions types to state history ABI.

P2P

• Normally wait for connection monitor to reconnect.
• Prevent sync ahead in irreversible mode.
• P2P: Allow irreversible mode syncing to continue when LIB is paused.
• Continue P2P connections while performing a snapshot.

Logging

• Fix reported time for produced blocks and disambiguate time in logs.
• Improve reported time for receiving and applying blocks.
• Update Produced block and Received block logs to report lib as the last irreversible block according to the reported received or produced block instead of the fork database root.
• Change handshake logs to debug instead of info.

System Compatibility

• Support Linux Kernel 4.15, by falling back to random_access_file for the older kernel.

System Stability

• Take into account that a claimed block may already be irreversible in consider_voting.

Tests

• Enhance test framework and test cases to sync and replay using actual committed reference blockchain data.
• Addition test coverage to validate version support for finalizer safety file.
• New test case where finality advances on lesser LIB.
• Split up long running tests so CICD can finish faster.
• Enable debug logging of the vote logger to support unittest.
• P2P: Increase timeout of sync throttle test.

Contributors

Special thanks to the contributors that submitted patches for this release:

• @linh2931
• @greg7mdp
• @heifner
• @spoonincode

Spring v1.0.1

This release introduces the V8 Snapshot format, which contains data missing from the V7 Snapshot format that may be needed to ensure continued syncing in some edge cases. Other changes include improvements to block production, replay, peer syncing, and deep-mind logging.

Note

Spring v1.0.1 cannot be loaded from V7 snapshots (created by Spring v1.0.0). It can load snapshot versions V6 and prior, as well as the new V8 snapshots. To upgrade a 1.0.0 node restarting from a compatible snapshot is required.

Notable Changes

• Introduced V8 Snapshots to fix potential loss of ability to sync in some edge cases due to missing data in V7 Snapshots
• Fixed bug in which block production could briefly pause after prior block producer missed their round
• Replay with --force-all-checks now validates quorum certificates (QCs) in the block extension
• P2P improvements to prevent unnecessary timeouts and redundant block requests
• Improved deep-mind logging

Complete Change Log

Bug Fixes

• Ensure header extension cache of blocks data loaded from data files is always available
• Validate block extensions during replay
• Introduce V8 Snapshot format
• Fix erroneous block production pause when prior block producer misses round

P2P Improvements

• Avoid redundant block requests
• Avoid timing out connection due to main thread being busy applying blocks

Deep-mind

• Use block's perspective of LIB instead of node's perspective of LIB in deep-mind logs
• Flush deep-mind logger after each message

Tests

• Test cases to validate edge cases related to missing V7 Snapshot data
• Rewrite P2P sync throttle test
• Improve test stability by preparing blocks for tests appropriately
• Add better test coverage of replay with --force-all-checks
• Correct reference fsi file used in tests
• Tests to verify expected behavior of replay of blocks with bad QC data

Other

• Verify block signature as soon as possible prior to any further block validation
• Consistent exception type on block extension validation
• Add earlier and more specific (yet backwards compatible) checks when applying policy diffs
• Cleanup error messages when upgrading from Spring v1.0.0
• Logging improvements for received blocks
• Utilize compiler to ensure blocks are treated as immutable after construction within implementation

Full Changelog: v1.0.0...v1.0.1

Contributors

Special thanks to the contributors that submitted patches for this release:

• @linh2931
• @heifner
• @greg7mdp

Spring v1.0.0

Antelope Spring is the successor to Leap as the C++ implementation of node software for blockchain networks based on the Antelope Protocol.

Spring 1.0.0 introduces Savanna consensus, which reduces time to finality from 3 minutes to 1 second under typical network conditions.

Also featured in Spring 1.0.0 is a significant reworking of the State History Plugin to improve performance, reliability, and support cross block chain consensus leveraging the faster finality of Savanna.

Note

To maintain IBC compatibility with EOS after it transitions to Savanna consensus, other Antelope chains will need to upgrade to a Savanna compatible IBC contract which depends on the activation of the BLS protocol feature that was first introduced in Leap v5.0.0.

Important Links

• Instructions to upgrade a node to Spring 1.0.0
• Instructions to transition a network to Savanna Consensus
• Recommendations for managing finalizer keys

New License Terms

Antelope Spring is made available under the terms of a Business Source License which is automatically granted for all usages which are Directly in Service of or Materially Dependent on the EOS Blockchain Network. Full details of the license terms may be found in the LICENSE file. Parties seeking a License for other usages may inquire by contacting [email protected].

Release Notes

Notable Changes

A subset of the changes in this release may require node operators and downstream service providers to make configuration changes. At a glance, these include:

• Introduced Savanna consensus
• Changes to some HTTP response codes
• Introduced new Snapshot format
• Disabled State Log History compression
• Introduced new State History configuration options
• Introduced new Finalizer configuration options
• Introduced new vote-threads option
• Introduced new max-reversible-blocks option

For actionable details and recommendations concerning the above changes, please see the Spring 1.0.0 upgrade guide. For complete technical details, consult the Complete Change Log and related pull requests below.

Important

The following features have been deprecated in Spring v1.0.0 and will be removed in Spring v2.0.0:

• The Prometheus plugin
• count field in get_table_by_scope response

Please see the Deprecation Log for more information.

Complete Change Log

Introduced in Spring v1.0.0

Savanna consensus

• Adds new chain_plugin option max-reversible-blocks. Specifies the maximum number of reversible blocks beyond current LIB to allow before the node is gracefully shutdown.
• New chain api get_finalizer_info with information on finalizers and last votes.
• Endpoint get_block_header_state updated for compatibility, and functionality reduced..
• Allow a larger number of proposer (producers) in Savanna from 125 to 64K. This matches the maximum number of allowed finalizers.
• Versioning for Finalizer Safety Information file.
• cleos support for get_finalizer_info.
• Create Savanna unittests modeled after the fast testnet wave tests
• Only vote on recent blocks and simply voting by comparing time.
• Switch to timestamp for better detection of most recent policy change.
• Use lastest_qc_claim block_num in block header.
• Add last_pending_finalizer_policy_generation into the finality digest calculation.
• Update logic to use latest timestamp to pick winning fork, previously used block height.
• Update finality digest adding level 3 commitments.
• Update logic to use latest timestamp to pick winning fork, previously used block height.
• Add timestamp and parent_timestamp to finality leaf nodes.
• Require a QC (quorum certificate) on both finalizer policy sets (active and pending) before considering a QC to have reached quorum on a block with a pending finalizer policy change.
• Switch to 2-chain for finality.
• For the convenience of finality violation proofs, make the finality digest to commit to the root digest of a Merkle tree over a sequence of records extracted from the sequence of block_refs in the block_refs. This Merkle root digest is called reversible_blocks_root.
• Revert previous changes to liveness rule. After evaluation the previous changes reduced liveness, and engineering identified a better solution.
• Write out chain_head after vote_processor threads have stopped.
• When a new finalizer policy or proposer policy is proposed in a block, the deep-mind logs should contain an entry associated with the block that contains the full policy (not just the diff). The keys within these policies should have string encoding.
• Move tracking of Savanna pending LIB into fork database. The fork database is in charge of tracking and persisting cached current pending Savanna LIB.
• Refactor ForkDB to consolidate concept of HEAD and better track the pending irreversible block.
• Validate QC by bitsize of votes.
• Clear out the state history log when Savanna transition is forked out.
• Improve accuracy of state history by selectively including finalizer policy.
• Update application of finalizer policies to avoid edge cases.
• When syncing/replaying optimize voting-start by processing on main thread.
• Extend scenarios where a deleted safety data file may be recovered.
• Avoid spamming peers with unnecessary votes when nodeos is first syncing old blocks.
• Align SHiP abi's by adding finality_data.
• Shift to new fork database only after irreversible transition is completed. Previously switch may occur before transition completed.
• Fix sporadic issues, update ForkDB to return block matching block number.
• Add strong QC block number to minimize size of proofs, and aid in finality violation proofs.
• Update controller to proper set producer authority under Savanna.
• Improve stability in P2P syncing for large forks with very small sync spans.
• Update to use the last pending finalizer policy in the history of the blockchain, if there are no pending finalizer policies in the queue, then the last pending finalizer policy is the active finalizer policy.
• Resolves duplicate polices from advancing. Finality can advance multiple blocks, previously this may cause duplicate policies to advance.
• Extend finality data to include full policy (including generation number) of the finalizer policy proposed in the associated block. This is used in SHiP data stream.
• Improve set_proposed_finalizers, has no side effects if transaction fails.
• Use the correct finalizer policy when logging.
• Avoid adding new proposer policy if there is no change.
• Improve set_finalizers host function to take a packed finalizer format.
• Move create and send vote off the main thread.
• Improved algorithm governing generation of finalizer policy.
• Avoid using fork in log messages that are not related to micro-forks.
• Check and test to ensure proposer policy is only set when there is a change.
• Update representation of variant formate for fc
• [Improve snapshot scheduling to avoid race conditions, and fixes issues where no snapshot was take...

Spring v1.0.0-rc3

Antelope Spring is the successor to Leap as the C++ implementation of node software for blockchain networks based on the Antelope Protocol.

Spring v1.0.0 introduces Savanna consensus, which reduces time to finality from 3 minutes to 1 second under typical network conditions.

Also featured in Spring 1.0.0 is a significant reworking of the State History Plugin to improve performance, reliability, and support cross block chain consensus leveraging the faster finality of Savanna.

Note

To maintain IBC compatibility with EOS after it transitions to Savanna consensus, other Antelope chains will need to upgrade to a Savanna compatible IBC contract which depends on the activation of the BLS protocol feature that was first introduced in Leap v5.0.0.

Important Links

• Instructions to upgrade a node to Spring 1.0.0
• Instructions to transition a network to Savanna Consensus
• Recommendations for managing finalizer keys

New License Terms

Antelope Spring is made available under the terms of a Business Source License which is automatically granted for all usages which are Directly in Service of or Materially Dependent on the EOS Blockchain Network. Full details of the license terms may be found in the LICENSE file. Parties seeking a License for other usages may inquire by contacting [email protected].

Release Notes

Notable Changes

Spring v1.0.0-rc3 fixes several defects reported against Spring v1.0.0-rc2. The most notable changes include:

• Improved liveness properties of consensus by modifying the rules that finalizers use to decide between voting strong vs weak
• Updated safety.dat format to include checksum to detect possible corruption on startup
• Modified behavior of max-reversible-blocks option to only pause production rather than shutdown
• Fixed regression in v1.0.0-rc2 that allowed the node to sync too far ahead
• Fixed crash when running SHiP during replay
• Additional improvements to logging and tests

Complete Change Log

Changes in Spring v1.0.0-rc3

Stability

• add BOOST_TEST_DEFAULTS_TO_CORE_DUMP to libtester users
• FSI serialization: support dynamically sized data and verify checksum
• Use other_branch_latest_time for liveness
• P2P: Prevent node from syncing too far ahead
• Pause production when max-reversible-blocks reached
• SHiP: Fix SHiP replay over savanna transition blocks
• Improve finalizer safety file exception handling

Tests

• Test verifying that proposer policy does not become active when finality has stalled
• Simplify disaster_recovery_2 test and verify correct LIB
• Test to demonstrate weak masking issue
• Improve restart production integration test
• Test validating new fsi design
• Fix liveness integration test flakiness
• Add integration tests for production-pause-vote-timeout

Maintainance

• various different signedness warning fixes
• bump chainbase submodule to main with signedness fix

Logging

• Additional debug output
• Logging improvements

Full Changelog: v1.0.0-rc2...v1.0.0-rc3

Contributors

Special thanks to the contributors that submitted patches for this release:

• @arhag
• @heifner
• @greg7mdp
• @spoonincode
• @linh2931

Spring v1.0.0-rc2

Antelope Spring is the successor to Leap as the C++ implementation of node software for blockchain networks based on the Antelope Protocol.

Spring v1.0.0 introduces Savanna consensus, which reduces time to finality from 3 minutes to 1 second under typical network conditions.

Also featured in Spring 1.0.0 is a significant reworking of the State History Plugin to improve performance, reliability, and support cross block chain consensus leveraging the faster finality of Savanna.

Note

To maintain IBC compatibility with EOS after it transitions to Savanna consensus, other Antelope chains will need to upgrade to a Savanna compatible IBC contract which depends on the activation of the BLS protocol feature that was first introduced in Leap v5.0.0.

Important Links

• Instructions to upgrade a node to Spring 1.0.0
• Instructions to transition a network to Savanna Consensus
• Recommendations for managing finalizer keys

New License Terms

Antelope Spring is made available under the terms of a Business Source License which is automatically granted for all usages which are Directly in Service of or Materially Dependent on the EOS Blockchain Network. Full details of the license terms may be found in the LICENSE file. Parties seeking a License for other usages may inquire by contacting [email protected].

Release Notes

Notable Changes

Spring v1.0.0-rc2 fixes several defects reported against Spring v1.0.0-rc1. The most notable changes include:

• The invalid snapshots generated in rc1 have been fixed in rc2.
• P2P changes were made fixing issues with frozen syncing observed in rc1

Important

The count field in the response to get_table_by_scope has been deprecated in this release and will be removed in Spring v2.0.0. See the Deprecation Log for more information.

Complete Change Log

Fixed in Spring v1.0.0-rc2

Snapshots

• v7 snapshot loading failures
• Loading from snapshot with block strides
• Check for chain_head.dat on snapshot load
• treat a end_block_num=0 as "forever" in snapshot scheduler for compatibility with leap 4.0

Syncing

• P2P: Sync not making progress after a timeout and reassign_fetch
• P2P: Sync mode stuck after hitting max-reversible-blocks, after restart
• P2P: Limit requested blocks to avoid max-reversible-blocks
• max-reversible-blocks syncing old blocks

Other

• log entry for legacy producer schedule broken in rc1
• Add changes required for updating to boost 1.86 - but without the boost upgrade
• Update chainbase to tip (initialization / move operations update)
• fix true_lowest() for uint128[2] index; possibly resolving some secondary table row RPC queries
• limit the vector size that can be reserved
• Remove unnecessary code that cannot be executed.
• Create Savanna unittests modeled after the fast testnet wave tests

Full Changelog: v1.0.0-rc1...v1.0.0-rc2

Contributors

Special thanks to the contributors that submitted patches for this release:

• @greg7mdp
• @heifner
• @spoonincode

Spring v1.0.0-rc1

Caution

This release (Spring v1.0.0-rc1) generates invalid snapshots and should not be used. Please use Spring v1.0.0-rc2.

Antelope Spring is the successor to Leap as the C++ implementation of node software for blockchain networks based on the Antelope Protocol.

Spring 1.0.0 introduces Savanna consensus, which reduces time to finality from 3 minutes to 1 second under typical network conditions.

Also featured in Spring 1.0.0 is a significant reworking of the State History Plugin to improve performance, reliability, and support cross block chain consensus leveraging the faster finality of Savanna.

Note

To maintain IBC compatibility with EOS after it transitions to Savanna consensus, other Antelope chains will need to upgrade to a Savanna compatible IBC contract which depends on the activation of the BLS protocol feature that was first introduced in Leap v5.0.0.

Important Links

• Instructions to upgrade a node to Spring 1.0.0
• Instructions to transition a network to Savanna Consensus
• Recommendations for managing finalizer keys

New License Terms

Antelope Spring is made available under the terms of a Business Source License which is automatically granted for all usages which are Directly in Service of or Materially Dependent on the EOS Blockchain Network. Full details of the license terms may be found in the LICENSE file. Parties seeking a License for other usages may inquire by contacting [email protected].

Release Notes

Notable Changes

A subset of the changes in this release may require node operators and downstream service providers to make configuration changes. At a glance, these include:

• Changes to some HTTP response codes
• Introduced new Snapshot format
• Disabled State Log History compression
• Introduced new State History configuration options
• Introduced new Finalizer configuration options
• Introduced new vote-threads option

For actionable details and recommendations concerning the above changes, please see the Spring 1.0.0 upgrade guide. For complete technical details, consult the Complete Change Log and related pull requests below.

Deprecation Notices

The count field in the response to get_table_by_scope has been deprecated in this release, and will be removed in the next major version of Spring. See the deprecation schedule for more details.

Complete Change Log

Introduced in Spring v1.0.0

Notice

Disable state history log file compression to improve performance. Users who would like to compress files can use other methods outside of nodeos.

Protocol

Adds new chain_plugin option max-reversible-blocks. Specifies the maximum number of reversible blocks beyond current LIB to allow before the node is gracefully shutdown.
New chain api get_finalizer_info with information on finalizers and last votes.
Endpoint get_block_header_state updated for compatibility, and functionality reduced..

Performance

Improve k1 signing performance ~10%, upgrade libsecp256k.
Instead of placing the entire producer proposed schedule in the instant_finality_extension, only put a diff of the producer proposed schedule in the extension.
Instead of placing the entire finalizer set in the instant_finality_extension, only put a diff of the active finalizer policy in the extension.
Allow a larger number of proposer (producers) in Savanna from 125 to 64K. This matches the maximum number of allowed finalizers.
Improve state history performance by using a buffer for the de/compression pipeline.

Internals

Guard against throwing an error write_deatastream's destructor.
P2P to correctly report id of know pending blocks.
Versioning for Finalizer Safety Information file.
P2P to correctly report id of know pending blocks.
Update libsecp256k1 faster speed and matches what Bitcoin Core is using.
Move block time check in speculative mode above watermark check in production mode.
Improve Syncing to handle condition where blocks were received before being applied. Previously this condition caused the node to not ever request the next range of blocks.
cleos support for get_finalizer_info.
More robust syncing to handle unavailable blocks, when client does not have backing blocklog.
Only vote on recent blocks and simply voting by comparing time.
Switch to timestamp for better detection of most recent policy change.
Use lastest_qc_claim block_num in block header.
Add last_pending_finalizer_policy_generation into the finality digest calculation.
Update logic to use latest timestamp to pick winning fork, previously used block height.
Update finality digest adding level 3 commitments.
Update logic to use latest timestamp to pick winning fork, previously used block height.
Add timestamp and parent_timestamp to finality leaf nodes.
Require a QC (quorum certificate) on both finalizer policy sets (active and pending) before considering a QC to have reached quorum on a block with a pending finalizer policy change.
Switch to 2-chain for finality.
For the convenience of finality violation proofs, make the finality digest to commit to the root digest of a Merkle tree over a sequence of records extracted from the sequence of block_refs in the block_refs. This Merkle root digest is called reversible_blocks_root.
Remove unneeded snapshot_block_state_legacy_v7.
Revert previous changes to liveness rule. After evaluation the previous changes reduced liveness, and engineering identified a better solution.
Write out chain_head after vote_processor threads have stopped.
Reduce Log Spam on shutdown.
When a new finalizer policy or proposer policy is proposed in a block, the deep-mind logs should contain an entry associated with the block that contains the full policy (not just the diff). The keys within these policies should have string encoding.
To support ease of use SHiP, finality_data captures a new pending finality policy, including encoded public keys from pending finality data.
Move tracking of Savanna pending LIB into fork database. The fork database is in charge of tracking and persisting cached current pending Savanna LIB.
Refactor ForkDB to consolidate concept of HEAD and better track the pending irreversible block.
Validate QC by bitsize of votes.
More robust block log replay through refactoring head block state management.
Add hardened mode for reproducible builds.
Upgrade to clang v18 in Docker repeatable build. Improvements to reduce Docker build cache.
Clear out the state history log when Savanna transition is forked out.
Improve accuracy of state history by selectively including finalizer policy.
Add cases where we update peer syncing status to more accurately reflect peer syncing from us status.
Improved block id duplication logic to retain the correct state history with replay/resync.
Update application of finalizer policies to avoid edge cases.
[More efficient P2P eliminating multiple requests for the same ran...