Explore the statistics: a query log to drill down by client, name, view and more; storage on the dashboard

jarld keeps every answer for a while (stats.retention.queries, default 7 days) in the statistics database: time, client, view, name, type, result and what blocked it. Every client, name, view, record type, result and blocklist on the Statistics page leads to an explore page for it: answers over time, the top values of every other dimension (each a link that narrows further) and the latest answers. GET /stats/explore.

The dashboard shows the size of Jarl's databases (and the query log's rows) and, per disk Jarl uses, how full it is and the databases' share.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

internal/stats/explore.go

@@ -0,0 +1,199 @@
+package stats
+
+import (
+	"context"
+	"fmt"
+	"strings"
+	"time"
+)
+
+// Exploring the query log (docs/stats.md#exploring): every answer of the
+// last days (stats.query_retention), filtered by any mix of client, name,
+// view, type, result and blocking, broken down by each of them.
+
+// ExploreFilter narrows the query log. Empty fields don't filter.
+type ExploreFilter struct {
+	Client string // a client key, as the top lists show it
+	Name   string // exactly this name
+	Domain string // this name or any below it
+	View   string
+	Type   string
+	RCode  string
+	// Blocked is "yes" (blocked by anything), "no", or a blocklist's name
+	// ("rules" for your own rules).
+	Blocked string
+	Since   time.Duration // how far back (default 24h)
+}
+
+// Dimensions are what the query log breaks down by, in the order shown.
+var Dimensions = []string{"client", "name", "view", "type", "rcode", "blocked"}
+
+// Exploration is what the query log says for a filter.
+type Exploration struct {
+	From  time.Time `json:"from"`
+	To    time.Time `json:"to"`
+	Total int64     `json:"total"`
+	// Blocked counts the blocked answers among them.
+	Blocked int64 `json:"blocked"`
+	// Series are the answers (and blocked ones) per Step seconds.
+	Step   int64          `json:"step"`
+	Series []ExplorePoint `json:"series"`
+	// By are the top values of each dimension the filter leaves open.
+	By map[string][]Entry `json:"by"`
+	// Recent are the latest answers.
+	Recent []QueryRow `json:"recent"`
+	// Oldest is the first answer the query log still has.
+	Oldest *time.Time `json:"oldest,omitempty"`
+	// Retention is how long it keeps them.
+	Retention string `json:"retention"`
+}
+
+// ExplorePoint is one step of the series.
+type ExplorePoint struct {
+	Time    int64 `json:"t"`
+	Count   int64 `json:"count"`
+	Blocked int64 `json:"blocked"`
+}
+
+// ErrNoQueryLog is returned when the query log is off.
+var ErrNoQueryLog = fmt.Errorf("the query log is off: it needs stats.query_log and a stats.query_retention above 0 in /etc/jarl/jarl.yaml (docs/stats.md#exploring)")
+
+// where builds the SQL condition and arguments for a filter.
+func (q *Querier) where(f ExploreFilter, from int64) (string, []any) {
+	conds := []string{"ts >= ?"}
+	args := []any{from}
+	eq := func(col, v string) {
+		if v != "" {
+			conds = append(conds, col+" = ?")
+			args = append(args, v)
+		}
+	}
+	eq("client", f.Client)
+	eq("name", strings.ToLower(strings.TrimSuffix(f.Name, ".")))
+	eq("view", f.View)
+	eq("type", strings.ToUpper(f.Type))
+	eq("rcode", strings.ToUpper(f.RCode))
+	if d := strings.ToLower(strings.TrimSuffix(f.Domain, ".")); d != "" {
+		conds = append(conds, "(name = ? OR (length(name) > length(?) AND substr(name, length(name) - length(?)) = ?))")
+		args = append(args, d, d, d, "."+d)
+	}
+	switch f.Blocked {
+	case "":
+	case "yes":
+		conds = append(conds, "blocked != ''")
+	case "no":
+		conds = append(conds, "blocked = ''")
+	default:
+		eq("blocked", f.Blocked)
+	}
+	if len(q.Views) > 0 {
+		marks := strings.TrimSuffix(strings.Repeat("?,", len(q.Views)), ",")
+		conds = append(conds, "view IN ("+marks+")")
+		for _, v := range q.Views {
+			args = append(args, v)
+		}
+	}
+	return strings.Join(conds, " AND "), args
+}
+
+// Explore answers a filter: totals, a series, the top values of each open
+// dimension (topN each) and the latest answers (recent of them).
+func (q *Querier) Explore(ctx context.Context, f ExploreFilter, topN, recent int) (*Exploration, error) {
+	if !q.Opts.QueryLog || q.Opts.RetainQueries <= 0 {
+		return nil, ErrNoQueryLog
+	}
+	now := q.Now()
+	since := f.Since
+	if since <= 0 {
+		since = 24 * time.Hour
+	}
+	if since > q.Opts.RetainQueries {
+		since = q.Opts.RetainQueries
+	}
+	from := now.Add(-since).Unix()
+	out := &Exploration{From: time.Unix(from, 0), To: now, By: map[string][]Entry{}, Series: []ExplorePoint{}, Recent: []QueryRow{},
+		Retention: humanDuration(q.Opts.RetainQueries)}
+	cond, args := q.where(f, from)
+	db := q.Store.db
+
+	if err := db.QueryRowContext(ctx, `SELECT count(*), coalesce(sum(blocked != ''), 0) FROM queries WHERE `+cond, args...).Scan(&out.Total, &out.Blocked); err != nil {
+		return nil, err
+	}
+	// The series: about 100 steps over the range, in whole minutes.
+	step := int64(since.Seconds()) / 100
+	step = max(60, step-step%60)
+	out.Step = step
+	rows, err := db.QueryContext(ctx, `SELECT ts - ts % ? AS t, count(*), sum(blocked != '') FROM queries WHERE `+cond+` GROUP BY t ORDER BY t`, append([]any{step}, args...)...)
+	if err != nil {
+		return nil, err
+	}
+	got := map[int64]ExplorePoint{}
+	for rows.Next() {
+		var p ExplorePoint
+		if err := rows.Scan(&p.Time, &p.Count, &p.Blocked); err != nil {
+			rows.Close()
+			return nil, err
+		}
+		got[p.Time] = p
+	}
+	rows.Close()
+	for t := from - from%step; t <= now.Unix(); t += step {
+		p := got[t]
+		p.Time = t
+		out.Series = append(out.Series, p)
+	}
+
+	open := map[string]bool{"client": f.Client == "", "name": f.Name == "", "view": f.View == "", "type": f.Type == "", "rcode": f.RCode == "", "blocked": f.Blocked == "" || f.Blocked == "yes"}
+	for _, dim := range Dimensions {
+		if !open[dim] {
+			continue
+		}
+		c := cond
+		if dim == "blocked" {
+			c += " AND blocked != ''"
+		}
+		rows, err := db.QueryContext(ctx, `SELECT `+dim+`, count(*) AS n FROM queries WHERE `+c+` GROUP BY `+dim+` ORDER BY n DESC, `+dim+` LIMIT ?`, append(args, topN)...)
+		if err != nil {
+			return nil, err
+		}
+		es := []Entry{}
+		for rows.Next() {
+			var e Entry
+			if err := rows.Scan(&e.Key, &e.Count); err != nil {
+				rows.Close()
+				return nil, err
+			}
+			es = append(es, e)
+		}
+		rows.Close()
+		out.By[dim] = es
+	}
+
+	rows, err = db.QueryContext(ctx, `SELECT ts, client, view, name, type, rcode, blocked FROM queries WHERE `+cond+` ORDER BY ts DESC, rowid DESC LIMIT ?`, append(args, recent)...)
+	if err != nil {
+		return nil, err
+	}
+	for rows.Next() {
+		var r QueryRow
+		if err := rows.Scan(&r.Time, &r.Client, &r.View, &r.Name, &r.Type, &r.RCode, &r.Blocked); err != nil {
+			rows.Close()
+			return nil, err
+		}
+		out.Recent = append(out.Recent, r)
+	}
+	rows.Close()
+
+	var oldest *int64
+	if err := db.QueryRowContext(ctx, `SELECT min(ts) FROM queries`).Scan(&oldest); err == nil && oldest != nil {
+		t := time.Unix(*oldest, 0)
+		out.Oldest = &t
+	}
+	return out, nil
+}
+
+// QueryLogSize returns how many answers the query log has.
+func (s *Store) QueryLogSize(ctx context.Context) (int64, error) {
+	var n int64
+	err := s.db.QueryRowContext(ctx, `SELECT count(*) FROM queries`).Scan(&n)
+	return n, err
+}