Backend
- web/service/metric_history.go: generic in-memory ring buffer with two
singletons — system-wide (cpu/mem/netUp/netDown/online/load1/5/15)
and per-node (cpu/mem) keyed by node id
- ServerService.AppendStatusSample writes all 8 metrics every 2s on the
same tick; AppendCpuSample/AggregateCpuHistory kept for back-compat
- NodeService.UpdateHeartbeat appends cpu/mem only on online ticks so
offline gaps render as missing data, not phantom dips
- New routes: GET /panel/api/server/history/:metric/:bucket and
GET /panel/api/nodes/history/:id/:metric/:bucket, both whitelisted
Frontend
- Sparkline component generalized: arbitrary value range (auto-scale
when valueMax=null), pluggable yFormatter/tooltipFormatter for B/s,
client counts, load averages
- SystemHistoryModal replaces CpuHistoryModal with tabs for every
metric; opened from a tag on the 3X-UI card next to Documentation
- NodeHistoryPanel: expandable row on the Nodes table showing per-node
CPU and Mem trends, refreshed every 15s
Localization
- Backfill systemHistoryTitle / trendLast2Min / pages.inbounds.{node,
deployTo, localPanel} and the entire pages.nodes block (51 keys
including statusValues + toasts) into all 11 non-en/fa locales:
ar-EG, es-ES, id-ID, ja-JP, pt-BR, ru-RU, tr-TR, uk-UA, vi-VN,
zh-CN, zh-TW
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
