Add live stats page with metrics sampling and route split

- New sandbox_metrics_snapshots table sampled every 10s (60-day retention) - Background MetricsSampler goroutine wired into control plane startup - GET /v1/sandboxes/stats?range=5m|1h|6h|24h|30d endpoint with adaptive polling intervals; reserved CPU/RAM uses ceil(paused/2) formula - StatsPanel component: 4 stat cards + 2 Chart.js line charts (straight lines, integer y-axis for running count, dual-axis for CPU/RAM) - Range filter persisted in URL query param; polls update data silently (no blink — loading state only shown on initial mount) - Split /dashboard/capsules into /list and /stats sub-routes with shared layout; capsuleRunningCount store syncs badge across routes - CreateCapsuleDialog extracted as reusable component
2026-03-25 14:41:05 +06:00
parent 2349f585ae
commit fee66bda50
21 changed files with 2059 additions and 1023 deletions
--- a/internal/db/metrics.sql.go
+++ b/internal/db/metrics.sql.go
@ -0,0 +1,141 @@
+// Code generated by sqlc. DO NOT EDIT.
+// versions:
+//   sqlc v1.30.0
+// source: metrics.sql
+
+package db
+
+import (
+	"context"
+
+	"github.com/jackc/pgx/v5/pgtype"
+)
+
+const getCurrentMetrics = `-- name: GetCurrentMetrics :one
+SELECT running_count, vcpus_reserved, memory_mb_reserved, sampled_at
+FROM sandbox_metrics_snapshots
+WHERE team_id = $1
+ORDER BY sampled_at DESC
+LIMIT 1
+`
+
+type GetCurrentMetricsRow struct {
+	RunningCount     int32              `json:"running_count"`
+	VcpusReserved    int32              `json:"vcpus_reserved"`
+	MemoryMbReserved int32              `json:"memory_mb_reserved"`
+	SampledAt        pgtype.Timestamptz `json:"sampled_at"`
+}
+
+func (q *Queries) GetCurrentMetrics(ctx context.Context, teamID string) (GetCurrentMetricsRow, error) {
+	row := q.db.QueryRow(ctx, getCurrentMetrics, teamID)
+	var i GetCurrentMetricsRow
+	err := row.Scan(
+		&i.RunningCount,
+		&i.VcpusReserved,
+		&i.MemoryMbReserved,
+		&i.SampledAt,
+	)
+	return i, err
+}
+
+const getPeakMetrics = `-- name: GetPeakMetrics :one
+SELECT
+    COALESCE(MAX(running_count), 0)::INTEGER      AS peak_running_count,
+    COALESCE(MAX(vcpus_reserved), 0)::INTEGER     AS peak_vcpus,
+    COALESCE(MAX(memory_mb_reserved), 0)::INTEGER AS peak_memory_mb
+FROM sandbox_metrics_snapshots
+WHERE team_id = $1
+  AND sampled_at > NOW() - INTERVAL '30 days'
+`
+
+type GetPeakMetricsRow struct {
+	PeakRunningCount int32 `json:"peak_running_count"`
+	PeakVcpus        int32 `json:"peak_vcpus"`
+	PeakMemoryMb     int32 `json:"peak_memory_mb"`
+}
+
+func (q *Queries) GetPeakMetrics(ctx context.Context, teamID string) (GetPeakMetricsRow, error) {
+	row := q.db.QueryRow(ctx, getPeakMetrics, teamID)
+	var i GetPeakMetricsRow
+	err := row.Scan(&i.PeakRunningCount, &i.PeakVcpus, &i.PeakMemoryMb)
+	return i, err
+}
+
+const insertMetricsSnapshot = `-- name: InsertMetricsSnapshot :exec
+INSERT INTO sandbox_metrics_snapshots (team_id, running_count, vcpus_reserved, memory_mb_reserved)
+VALUES ($1, $2, $3, $4)
+`
+
+type InsertMetricsSnapshotParams struct {
+	TeamID           string `json:"team_id"`
+	RunningCount     int32  `json:"running_count"`
+	VcpusReserved    int32  `json:"vcpus_reserved"`
+	MemoryMbReserved int32  `json:"memory_mb_reserved"`
+}
+
+func (q *Queries) InsertMetricsSnapshot(ctx context.Context, arg InsertMetricsSnapshotParams) error {
+	_, err := q.db.Exec(ctx, insertMetricsSnapshot,
+		arg.TeamID,
+		arg.RunningCount,
+		arg.VcpusReserved,
+		arg.MemoryMbReserved,
+	)
+	return err
+}
+
+const pruneOldMetrics = `-- name: PruneOldMetrics :exec
+DELETE FROM sandbox_metrics_snapshots
+WHERE sampled_at < NOW() - INTERVAL '60 days'
+`
+
+func (q *Queries) PruneOldMetrics(ctx context.Context) error {
+	_, err := q.db.Exec(ctx, pruneOldMetrics)
+	return err
+}
+
+const sampleSandboxMetrics = `-- name: SampleSandboxMetrics :many
+SELECT
+    team_id,
+    (COUNT(*) FILTER (WHERE status IN ('running', 'starting')))::INTEGER                                              AS running_count,
+    (COALESCE(SUM(vcpus)    FILTER (WHERE status IN ('running', 'starting')), 0)
+     + CEIL(COALESCE(SUM(vcpus)    FILTER (WHERE status = 'paused'), 0)::NUMERIC / 2))::INTEGER                      AS vcpus_reserved,
+    (COALESCE(SUM(memory_mb) FILTER (WHERE status IN ('running', 'starting')), 0)
+     + CEIL(COALESCE(SUM(memory_mb) FILTER (WHERE status = 'paused'), 0)::NUMERIC / 2))::INTEGER                     AS memory_mb_reserved
+FROM sandboxes
+WHERE status IN ('running', 'starting', 'paused')
+GROUP BY team_id
+`
+
+type SampleSandboxMetricsRow struct {
+	TeamID           string `json:"team_id"`
+	RunningCount     int32  `json:"running_count"`
+	VcpusReserved    int32  `json:"vcpus_reserved"`
+	MemoryMbReserved int32  `json:"memory_mb_reserved"`
+}
+
+// Aggregates per-team resource usage from the live sandboxes table.
+// paused sandboxes count at 50% (ceil) for capacity reservation.
+func (q *Queries) SampleSandboxMetrics(ctx context.Context) ([]SampleSandboxMetricsRow, error) {
+	rows, err := q.db.Query(ctx, sampleSandboxMetrics)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+	var items []SampleSandboxMetricsRow
+	for rows.Next() {
+		var i SampleSandboxMetricsRow
+		if err := rows.Scan(
+			&i.TeamID,
+			&i.RunningCount,
+			&i.VcpusReserved,
+			&i.MemoryMbReserved,
+		); err != nil {
+			return nil, err
+		}
+		items = append(items, i)
+	}
+	if err := rows.Err(); err != nil {
+		return nil, err
+	}
+	return items, nil
+}