fix: raise MySQL pool 8->16 — 8 starved the instance's own live-sync polling (revision + grid refresh + live_status + chat), so the grid stopped showing other operators' QSOs; 16 avoids the stall and still fits ~15 ops under max_connections=300

internal/db/mysql.go

@@ -129,19 +129,21 @@ func translateTextColumns(s string) string {
 // OpenMySQL opens the shared MySQL logbook, creating the database if needed,
 // then applies the (translated) embedded migrations. multiStatements is enabled
 // so multi-statement migration files run in a single Exec.
-// tuneMySQLPool sizes the connection pool for a SHARED server. A multi-operator
+// tuneMySQLPool sizes the connection pool for a SHARED server. Two opposing needs:
-// special event runs many OpsLog instances against ONE MySQL, whose max_connections
+//   - A big per-instance pool (we shipped 50) let a handful of ops exhaust the
-// is a hard global limit (often 100–151). A big per-instance pool (we used 50) let
+//     server's global max_connections — "Error 1040: Too many connections".
-// a handful of ops exhaust the server — "Error 1040: Too many connections" — after
+//   - Too SMALL a pool starves this instance's OWN concurrent UI queries — the live
-// which queries (the on-air widget, award_refs, the grid) fail silently and the UI
+//     multi-op sync (revision poll every 2s + a 3-query grid refresh), live_status,
-// looks frozen. A modest cap with brisk idle reaping keeps each instance's server
+//     chat, stats, cluster — so the grid stops showing other operators' QSOs. (8 was
-// footprint small so ~10+ ops fit, while still absorbing the UI's concurrent
+//     too tight and reintroduced this "deadlock-like stall".)
-// logbook queries. No max lifetime: a slow first migration can run for minutes on a
+// 16 is the balance: plenty for the UI's bursts, and ~15 ops still fit under a
-// single connection and reaping it mid-migration drops the selected database.
+// raised max_connections=300 (advise operators to raise it for a big multi-op).
+// Brisk idle reaping returns slots to the server; no max lifetime so a slow first
+// migration on one connection isn't reaped mid-run (drops the selected database).
 func tuneMySQLPool(conn *sql.DB) {
-	conn.SetMaxOpenConns(8)
+	conn.SetMaxOpenConns(16)
-	conn.SetMaxIdleConns(3)
+	conn.SetMaxIdleConns(6)
-	conn.SetConnMaxIdleTime(30 * time.Second) // return idle connections to the server promptly
+	conn.SetConnMaxIdleTime(30 * time.Second)
 	conn.SetConnMaxLifetime(0)
 }