mosona-labs/mosona-manager v0.1.14 on GitHub

Please refer to the Quickstart (Docs) for detailed instructions on installation and configuration.

Full Changelog: v0.1.13...v0.1.14

Logs API no longer supports offset pagination. GET /api/v1/logs and GET /api/admin/logs reject page greater than 1 with HTTP 400 (Offset pagination is no longer supported). Responses return next_cursor and has_more instead of total. The bundled web UI is updated; third-party clients must follow next_cursor (omit it for the newest page). page=1 remains accepted as an alias for the first page.
Log queries default to the last 30 days. Callers may pass RFC3339 start and end (maximum span 365 days, unchanged). Message search is limited to a 30-day window; a longer range with message set returns 400. Older points remain in InfluxDB untouched.

Optional RFC3339 start/end filters on team and admin log list endpoints, with cursor-based paging (cursor, next_cursor, has_more).

Share one monitor snapshot per team across SSE subscribers (refresh every 3s, 8s query timeout, 64 concurrent Influx loads) instead of querying InfluxDB once per open dashboard.
Queue and batch server-status writes to InfluxDB (10k-point queue, 500-point batches, 1s flush), retry a failed batch once, drop the oldest points on overflow, and drain the queue on shutdown alongside audit logs.
Batch alert observation queries (64 servers at a time, grouped by metric and for_duration) and pre-aggregate windows instead of one Influx query per server/rule.
Parallelize admin-dashboard Influx queries (record counts and system usage) with a 15s context timeout.

Reject parent path segments (..) in SafeJoinUnderRoot, including backslash-separated paths, so static-file joins cannot walk via traversal segments that previously cleaned back under the root.
Validate admin list pagination (page/size) against shared bounds (default 1/20, max 100000/1000) and return 400 on invalid values instead of coercing them.
Escape LIKE wildcards in admin user/team search, match numeric IDs exactly, and list teams without requiring a member join so empty teams appear in the admin list.

Virtualize dashboard server cards and throttle SSE snapshot commits so large fleets stay interactive while status updates stream.
Replace log page numbers with previous/next cursor paging and a time-range selector (24h / 7d / 30d / 90d / 365d); message search clamps the range to 30 days.
Render agent-mode servers on the terminal page when SSH/OS fields are absent instead of throwing on null address/username.
Correct monitor chart downsampling so longer ranges (7d / 30d / 180d / 365d) keep a sensible number of points instead of collapsing or over-aggregating.