v1.4.6

Changed

• Improved reconciliation after loss of local database
• Added detection for multiple agents competing over the same device

v1.4.5

Changed

• Added support for AWS_SESSION_TOKEN
• Fixed an issue with HTTP/2 keepalive timeout configuration
• Changed log level for messages about empty bag files, from info to debug

v1.4.4

Changed

• Enabled HTTP/2 keepalives for uploads, where available
• Switched to a single-threaded runtime
• Added release version to user-agent string

v1.4.3

Changed

• Switched to WAL journal mode for sqlite database, which should improve performance.
• Gracefully cancel pending uploads in response to shutdown signals.

Fixed

• Fixed a bug where the agent would report "database locked" errors due to disk contention.
• Fixed a bug where the agent would not respond to shutdown signals in a timely fashion.
• Fixed a bug where the agent would not retry failed uploads for recordings selected by the WATCH_AUTO_IMPORT_PATTERN configuration.

v1.4.2

Added

• Added WATCH_USE_NOTIFY, which can be used to toggle filesystem notifications.

Fixed

• Fixed a bug introduced in v1.3.0, where the agent filesystem watcher may deadlock and fail to pick up new recordings.

v1.4.1

Fixed

• Fixed a bug introduced in v1.3.3, where the agent would time out data uploads to foxglove-hosted storage.

v1.4.0

Changed

• Switched from OpenSSL to rustls.
• Removed the runtime dependency on glibc by statically linking the agent binary against musl.

v1.3.3

Fixed

• Fixed an issue where API calls would stall indefinitely, blocking submission of new recordings

v1.3.2

Feature

• Added configuration option (--multipart-part-size/MULTIPART_PART_SIZE) for multipart upload part size, allowing larger file uploads.

v1.3.1

Fixed

• Fixed an issue where the agent fails to insert recording topics after upgrading to 1.3.0.