lxd/api_metrics: Filter query for Warnings appropriately

Filters query for Warnings on the metrics handler by Node. Since some
Warnings do not have a node, nodeless Warnings are only being included
if querying the metrics from the leader node.

Signed-off-by: hamistao <[email protected]>

lxd/api_metrics.go

@@ -14,6 +14,7 @@ import (
 	"github.com/canonical/lxd/lxd/auth"
 	"github.com/canonical/lxd/lxd/db"
 	dbCluster "github.com/canonical/lxd/lxd/db/cluster"
+	"github.com/canonical/lxd/lxd/db/warningtype"
 	"github.com/canonical/lxd/lxd/instance"
 	instanceDrivers "github.com/canonical/lxd/lxd/instance/drivers"
 	"github.com/canonical/lxd/lxd/instance/instancetype"
@@ -129,7 +130,7 @@ func metricsGet(d *Daemon, r *http.Request) response.Response {
 		}
 
 		// Register internal metrics.
-		intMetrics = internalMetrics(ctx, s.StartTime, tx)
+		intMetrics = internalMetrics(ctx, s, tx)
 		return nil
 	})
 	if err != nil {
@@ -378,10 +379,39 @@ func getFilteredMetrics(s *state.State, r *http.Request, compress bool, metricSe
 	return response.SyncResponsePlain(true, compress, metricSet.String())
 }
 
-func internalMetrics(ctx context.Context, daemonStartTime time.Time, tx *db.ClusterTx) *metrics.MetricSet {
+// clusterMemberWarnings returns the list of unresolved warnings related to this cluster member.
+// If this member is the leader, also include nodeless warnings.
+// This way we include them while avoiding counting them redundantly across cluster members.
+func clusterMemberWarnings(ctx context.Context, s *state.State, tx *db.ClusterTx) ([]dbCluster.Warning, error) {
+	var filters []dbCluster.WarningFilter
+
+	leaderInfo, err := s.LeaderInfo()
+	if err != nil {
+		return nil, err
+	}
+
+	// Use local variable to get pointer.
+	emptyNode := ""
+
+	for status := range warningtype.Statuses {
+		// Do not include resolved warnings that are resolved but not yet pruned.
+		if status != warningtype.StatusResolved {
+			filters = append(filters, dbCluster.WarningFilter{Node: &s.ServerName, Status: &status})
+			if leaderInfo.Leader {
+				// Count the nodeless warnings as belonging to the leader node.
+				filters = append(filters, dbCluster.WarningFilter{Node: &emptyNode, Status: &status})
+			}
+		}
+	}
+
+	return dbCluster.GetWarnings(ctx, tx.Tx(), filters...)
+}
+
+func internalMetrics(ctx context.Context, s *state.State, tx *db.ClusterTx) *metrics.MetricSet {
 	out := metrics.NewMetricSet(nil)
 
-	warnings, err := dbCluster.GetWarnings(ctx, tx.Tx())
+	warnings, err := clusterMemberWarnings(ctx, s, tx)
+
 	if err != nil {
 		logger.Warn("Failed to get warnings", logger.Ctx{"err": err})
 	} else {
@@ -421,7 +451,7 @@ func internalMetrics(ctx context.Context, daemonStartTime time.Time, tx *db.Clus
 	}
 
 	// Daemon uptime
-	out.AddSamples(metrics.UptimeSeconds, metrics.Sample{Value: time.Since(daemonStartTime).Seconds()})
+	out.AddSamples(metrics.UptimeSeconds, metrics.Sample{Value: time.Since(s.StartTime).Seconds()})
 
 	// Number of goroutines
 	out.AddSamples(metrics.GoGoroutines, metrics.Sample{Value: float64(runtime.NumGoroutine())})