Releases: stellar/stellar-core
Releases · stellar/stellar-core
v12.0.0rc2
Stability improvements
- Fixed a slow leak that would cause buckets to not be deleted from disk until the next restart of stellar-core
- Change the summary quorum information in the info endpoint to use the last SCP round. This makes it easier to use that field for monitoring.
- Fix a few logging issues in overlay.
- Fix a race in Windows builds that could cause intermittent failures during catchup.
- Fix a mismatch in invariant names with the example configuration.
New features
- See protocol changes below.
- The
peersendpoint now includes how long the connection has been active
Breaking changes
- Protocol 12 support
v12.0.0rc1
Release notes:
This release adds support for protocol 12 as well as regular improvements.
Stability improvements
- Change the summary quorum information in the info endpoint to use the last SCP round. This makes it easier to use that field for monitoring.
- Fix a few logging issues in overlay.
- Fix a race in Windows builds that could cause intermittent failures during catchup.
- Fix a mismatch in invariant names with the example configuration.
New features
- See protocol changes below.
- The
peersendpoint now includes how long the connection has been active
Breaking changes
v11.4.0
Release notes:
Stability improvements
- Reduce disk utilization when catching up by not downloading everything before performing actual work.
- Force sync bucket files when performing bucket operation. There was a small chance that buckets would be “lost” during certain events like a hard reboot, causing the instance to be in an unrecoverable state.
- Do not try to “fix missing buckets” as this is misleading
- Fixed race condition (caused by doing redundant work) when publishing to multiple archives
- Fixed http command to ban nodes that was broken in a previous release.
- Automated quorum set generator was generating invalid quorum sets when using all 3 quality levels in a single configuration.
- Add a way for load generator to mark run as “failed”.
- Fix a race condition on Windows that was causing catchup to fail.
Breaking changes
- Squash SQL schema changes, it’s now not possible to upgrade an instance older than 11.0.0
v11.4.0rc2
Release notes:
Stability improvements
- Reduce disk utilization when catching up by not downloading everything before performing actual work.
- Force sync bucket files when performing bucket operation. There was a small chance that buckets would be “lost” during certain events like a hard reboot, causing the instance to be in an unrecoverable state.
- Do not try to “fix missing buckets” as this is misleading
- Fixed race condition (caused by doing redundant work) when publishing to multiple archives
- Fixed http command to ban nodes that was broken in a previous release.
- Automated quorum set generator was generating invalid quorum sets when using all 3 quality levels in a single configuration.
- Add a way for load generator to mark run as “failed”.
Breaking changes
- Squash SQL schema changes, it’s now not possible to upgrade an instance older than 11.0.0
v11.4.0rc1
Release notes:
Stability improvements
- Reduce disk utilization when catching up by not downloading everything before performing actual work.
- Force sync bucket files when performing bucket operation. There was a small chance that buckets would be “lost” during certain events like a hard reboot, causing the instance to be in an unrecoverable state.
- Do not try to “fix missing buckets” as this is misleading
- Fixed race condition (caused by doing redundant work) when publishing to multiple archives
- Fixed http command to ban nodes that was broken in a previous release.
- Automated quorum set generator was generating invalid quorum sets when using all 3 quality levels in a single configuration.
- Add a way for load generator to mark run as “failed”.
Breaking changes
- Squash SQL schema changes, it’s now not possible to upgrade an instance older than 11.0.0
v11.3.0
Release notes:
Stability improvements
- fixed a couple issues where stellar-core would perform a lot of redundant bucket merges when publishing to history
- specifying
fullkeys=trueto commands such asquorumnow has a consistent behavior, in that when not specified aliases or short keys are returned and when specified only full keys are returned - fixes an issue where in some rare cases transactions for certain accounts would be rejected due to the banning/invalidation of the first transaction in a sequence of transactions for the same account
- fixed an issue that was causing command line catchup to panic in some cases
- make transitive quorum checking recomputation more deterministic
New features
- core now computes a new quorum intersection criticality which allows to detect potential future quorum intersection problems
- added new metrics to track performance of various subsystems
- ability to replay history with modified bucket list logic (testing)
v11.3.0rc1
Release notes:
Stability improvements
- fixed a couple issues where stellar-core would perform a lot of redundant bucket merges when publishing to history
- specifying
fullkeys=trueto commands such asquorumnow has a consistent behavior, in that when not specified aliases or short keys are returned and when specified only full keys are returned - fixes an issue where in some rare cases transactions for certain accounts would be rejected due to the banning/invalidation of the first transaction in a sequence of transactions for the same account
- fixed an issue that was causing command line catchup to panic in some cases
- make transitive quorum checking recomputation more deterministic
New features
- core now computes a new quorum intersection criticality which allows to detect potential future quorum intersection problems
- added new metrics to track performance of various subsystems
- ability to replay history with modified bucket list logic (testing)
v11.2.0
Stability Improvements
- Do not allow gaps to form in history, stellar-core will always replay transactions instead of “jumping forward” post the initial sync. This gets rid of a common issue with Horizon.
- Do not corrupt local node on unknown upgrade.
- SCP messages are persisted across multiple ledgers. One of the root causes for May’s outage.
- Global log level can be set again.
New Features
- Quorum intersection monitoring: Allows administrators to set up monitoring via the logs when the overall network is not configured properly for quorum intersection.
- Automatic Quorum Set generation: Allows administrators to automatically configure their quorum set according to best practices. This is now the preferred way for admins to configure their quorum sets (largely motivated by May’s outages).
- Added new metrics to track SCP timeouts.
For more information on the motivation and implementation of the features related to quorum intersection, see our recent blog post on the developer blog.
Breaking Changes
- The
infoendpoint changed in a breaking way (specifically, thequorumsection - see observing consensus). - The old style command line syntax has been removed after being deprecated since November 2018.
- Windows builds now require Visual Studio 2019.
v11.2.0rc3
Stability improvements
- Do not allow gaps to form in history, stellar-core will always replay transactions instead of “jumping forward” post the initial sync. This gets rid of a common issue with Horizon.
- Do not corrupt local node on unknown upgrade.
- SCP messages are persisted across multiple ledgers. One of the root causes for May’s outage.
- Global log level can be set again
New features
- Quorum intersection monitoring ; Allows to alarm when the overall network is not configured properly
- Automatic Quorum Set generation ; The now preferred way for configuring quorum sets (mitigation for May’s outages)
- Added new metrics to track SCP timeouts
Breaking changes
- The old style command line has been removed ; was deprecated since November
- Windows builds now require Visual Studio 2019
v11.2.0rc2
Stability improvements
- Do not allow gaps to form in history, stellar-core will always replay transactions instead of “jumping forward” post the initial sync. This gets rid of a common issue with Horizon.
- Do not corrupt local node on unknown upgrade.
- SCP messages are persisted across multiple ledgers. One of the root causes for May’s outage.
- Global log level can be set again
New features
- Quorum intersection monitoring ; Allows to alarm when the overall network is not configured properly
- Automatic Quorum Set generation ; The now preferred way for configuring quorum sets (mitigation for May’s outages)
- Added new metrics to track SCP timeouts
Breaking changes
- The old style command line has been removed ; was deprecated since November
- Windows builds now require Visual Studio 2019