Releases: benbjohnson/litestream
v0.3.9
This release fixes Docker builds for ARM as well as upgrades dependencies such as LZ4 to fix known bugs.
v0.3.8
This release upgrades the version of SQLite to 3.38.0.
v0.3.8-beta0
This release simply upgrades the SQLite version to v3.38.0.
v0.3.7
This release fixes an issue where the wrong snapshot can be removed when multiple generations are present.
v0.3.6
Fixes an issue where LIST
operations are performed too often and increase the cost of running Litestream.
v0.4.0-alpha.3
Refactor shadow WAL to use segments
v0.3.5
This release focused on refactoring the replica code to more easily support new replica types. We've added Azure Blob Storage (abs
), Google Cloud Storage (gcs
), and SFTP (sftp
) replica types.
We've also included support for simple process supervision by adding an -exec
flag to litestream replicate
so folks don't need to use s6
when running in a Docker container. Kubernetes support has improved with a new -if-db-not-exists
flag on litestream restore
. It allows users to run autorecovery using the official Litestream Docker image.
The S3 sync interval has been lowered to 1s
from 10s
. It can increase cost if your application continually writes to the database, however, most applications do not so this reduces the window for data loss. We've added support for older versions of Linux such as Ubuntu 18.04, added built-in support for Filebase replica URLs, and updated replica metrics.
You can find a full description of changes in the release notes on the blog: https://litestream.io/blog/litestream-v0.3.5-released/
v0.3.4
Overview
The v0.3.4 release focused on Docker & ARM support as well as updating to a more permissive license. You can find a detailed post about all the changes on the Litestream blog:
https://litestream.io/blog/litestream-v0.3.4-released/
Changelog
- Updated license from GPLv3 to APLv2.
- Add Docker image
- Reduce binary size
- Add ARM support
- Add
-if-replica-exists
flag to restore - Add static release builds
- Sync to replicas on close
- Remove SQLite write lock during WAL sync
- Allow use of LITESTREAM prefixed environment variables
- Improve restore performance
- Fix snapshot selection bug during restore-by-index
- Use hex-encoding for restore's index flag
- Improve restoration logging
- Remove -dry-run flag in restore
- Use microsecond resolution for logging
- Ensure minimum wait time for TestDB_UpdatedAt/WAL test
- Add osusergo & netgo tags for static builds
- Omit load extensions for static builds
- Configuration file environment variable expansion
- Remove reference to "wal" in first db init command
- Add skip-verify flag for using self-signed certificates
- Catch sigterm & add shutdown logging
v0.3.3
Overview
This release was focused on adding Windows Services support, adding S3-compatible store support, and a variety of quality-of-life improvements.
Changelog
- Add Windows Service & MSI builds (#65)
- Add support for S3-compatible stores. All stores supported but added replica URL support for MinIO, Backblaze B2, & GCS. (#66)
- Fix bug with propagation of global config settings (#70)
- Fix format of
etc/litestream.yml
example file (#73) - Gracefully handle database initialization failures and retry later (#82)
- Add snapshot interval to allow multiple snapshots per retention period (#84)
- Fix locking issue on operating systems that don't support OFD locks (#93)
- Prevent user from specifying both a config file & a replica URL (#94)
- Add acknowledgements for folks who contribute their time and effort to help test, debug, and provide feedback! (#96)
- Fix reported metrics URL when specifying a
host:port
(#103) - Expose additional database field via configuration (#105)
- Add additional debug information on WAL header mismatch during restart (#109)
- Add replica URL support for DigitalOcean Spaces (#110)
- Add replica URL support for Linode Object Storage (#111)
- Default region when endpoint is set (#118)
- Default to use path style when endpoint is set (#120)
v0.3.2
This release makes several small quality-of-life improvements and fixes a shadow WAL corruption bug that can occur under certain circumstances with large databases.
- Fix tabwriter to align using spaces instead of tabs.
- Reduce s3 sync interval when using replica URL to improve new user experience.
- Add trace file to
replicate
command for debugging. - Fix shadow WAL corruption on stalled validation.
- Enforce max WAL index to prevent theoretical overflow.
- Check checkpoint result during restore.
- Fix Windows build (although not supported yet).
- Log WAL frame checksum mismatch.
- Reduce logging output.