Release Notes

v1.10.4

Added an alert system that allows users to configure alerts via the web interface.

Fixed a missing Content-Type: application/json header in fetch task callback requests.

Added retry functionality for HTTP requests with incorrect body lengths.
Introduced error rate monitoring for nodes and real-time transfer speed display.
Added node disconnection alerts, with disconnected nodes marked in red on the web interface.

Fixed a bug where only the top-level prefix was validated during prefix auto-discovery listing.
Ensured that empty directories in the source are synchronized to the destination when prefix auto-discovery is disabled.
Added support for Qiniu private cloud endpoints.

Optimized data statistics collection.
Resolved a panic issue caused by empty nextmarker during S3 list fetching.
Fixed upload ID invalidation due to canceled multipart upload requests during retries.
Resolved an issue with list operations always using the same marker.
Added functionality to set usernames and passwords during the first login if not configured at startup.

Introduced fetch task mode, allowing incremental file synchronization via POST requests to the /fetch endpoint.
Added support for specifying the --sentinel-addr parameter at startup for connecting nodes to the Pika database in sentinel mode.
Enabled HTTPS downloads and list fetching for AWS, US3, OSS, Qiniu, and URL list endpoint types.
Added automatic prefix discovery for endpoint list fetching, enabling parallel listing for each prefix.
Introduced the prefix-file-list API for US3 endpoints, useful in private cloud environments.
Enhanced multipart upload functionality with dynamic chunk size based on initialization responses.
Treated source 404 errors as task failures.
Implemented an exponential backoff strategy for worker heartbeats when disconnected from the master, with a maximum interval of 32 seconds.
Ensured proper abortion of incomplete multipart uploads by worker nodes.
Improved handling of HTTP status codes (404, 503, 500, 504) during downloads with exponential backoff retries (up to 5 times).
Optimized task allocation for quicker recovery after worker failures.
Enhanced the web interface to display file list operation progress during task listing and auditing.
Added regex-based filtering for source object keys during task creation.

Introduced sync deletion, allowing users to remove files from the destination that do not exist in the source.
Improved the web interface by separating synchronization, deletion, and active metric displays.