Avoid duplicate inserts to preserve DB space

ammario · ammario · commit 9bd442cfaf43 · 2022-08-30T23:19:54.000Z
diff --git a/agent/agent.go b/agent/agent.go
@@ -351,9 +351,7 @@ func (a *agent) init(ctx context.Context) {
 	go a.run(ctx)
 	if a.statsReporter != nil {
 		cl, err := a.statsReporter(ctx, a.logger, func() *Stats {
-			ss := a.stats.Copy()
-			a.stats.Reset()
-			return ss
+			return a.stats.Copy()
 		})
 		if err != nil {
 			a.logger.Error(ctx, "report stats", slog.Error(err))
diff --git a/agent/stats.go b/agent/stats.go
@@ -63,17 +63,6 @@ func (s *Stats) Copy() *Stats {
 	return &ss
 }
 
-func (s *Stats) Reset() {
-	s.Lock()
-	defer s.Unlock()
-
-	for _, ps := range s.ProtocolStats {
-		atomic.StoreInt64(&ps.NumConns, 0)
-		atomic.StoreInt64(&ps.RxBytes, 0)
-		atomic.StoreInt64(&ps.TxBytes, 0)
-	}
-}
-
 // goConn launches a new connection-processing goroutine, account for
 // s.Conns in a thread-safe manner.
 func (s *Stats) goConn(conn net.Conn, protocol string, fn func(conn net.Conn)) {
@@ -91,15 +80,12 @@ func (s *Stats) goConn(conn net.Conn, protocol string, fn func(conn net.Conn)) {
 	}
 
 	atomic.AddInt64(&ps.NumConns, 1)
-	go fn(cs)
+	go func() {
+		fn(cs)
+	}()
 }
 
 // StatsReporter periodically accept and records agent stats.
-// The agent should send incremental stats instead of the cumulative
-// value so that SQL queries can efficiently detect activity rates and
-// short-lived connections.
-//
-// E.g., we want to easily query for periods where transfers exceeded 100MB.
 type StatsReporter func(
 	ctx context.Context,
 	log slog.Logger,
diff --git a/coderd/metrics.go b/coderd/metrics.go
@@ -4,6 +4,7 @@ import (
 	"encoding/json"
 	"net/http"
 	"os"
+	"reflect"
 	"strconv"
 	"time"
 
@@ -19,8 +20,6 @@ import (
 	"github.com/coder/coder/codersdk"
 )
 
-const AgentStatIntervalEnv = "CODER_AGENT_STAT_INTERVAL"
-
 func FillEmptyDAUDays(rows []database.GetDAUsFromAgentStatsRow) []database.GetDAUsFromAgentStatsRow {
 	var newRows []database.GetDAUsFromAgentStatsRow
 
@@ -77,6 +76,8 @@ func (api *API) daus(rw http.ResponseWriter, r *http.Request) {
 	httpapi.Write(rw, http.StatusOK, resp)
 }
 
+const AgentStatIntervalEnv = "CODER_AGENT_STAT_INTERVAL"
+
 func (api *API) workspaceAgentReportStats(rw http.ResponseWriter, r *http.Request) {
 	api.websocketWaitMutex.Lock()
 	api.websocketWaitGroup.Add(1)
@@ -123,6 +124,7 @@ func (api *API) workspaceAgentReportStats(rw http.ResponseWriter, r *http.Reques
 	}
 	defer conn.Close(websocket.StatusAbnormalClosure, "")
 
+	// The complexity of the DAU query is inversely proportional to this interval.
 	var interval = time.Minute
 
 	// Allow overriding the stat interval for debugging and testing purposes.
@@ -141,6 +143,7 @@ func (api *API) workspaceAgentReportStats(rw http.ResponseWriter, r *http.Reques
 
 	ctx := r.Context()
 	timer := time.NewTicker(interval)
+	var lastReport codersdk.AgentStatsReportResponse
 	for {
 		err := wsjson.Write(ctx, conn, codersdk.AgentStatsReportRequest{})
 		if err != nil {
@@ -170,15 +173,20 @@ func (api *API) workspaceAgentReportStats(rw http.ResponseWriter, r *http.Reques
 			return
 		}
 
+		// Avoid inserting duplicate rows to preserve DB space.
+		var insert = !reflect.DeepEqual(lastReport, rep)
+
 		api.Logger.Debug(ctx, "read stats report",
 			slog.F("agent", workspaceAgent.ID),
 			slog.F("resource", resource.ID),
 			slog.F("workspace", workspace.ID),
+			slog.F("insert", insert),
 			slog.F("payload", rep),
 		)
 
-		// Avoid inserting empty rows to preserve DB space.
-		if len(rep.ProtocolStats) > 0 {
+		if insert {
+			lastReport = rep
+
 			_, err = api.Database.InsertAgentStat(ctx, database.InsertAgentStatParams{
 				ID:          uuid.NewString(),
 				CreatedAt:   time.Now(),
