migrate changes from cc-backend PR#364

internal/metricstoreclient/cc-metric-store-queries.go

@@ -70,14 +70,15 @@ func (ccms *CCMetricStore) buildQueries(
 	scopes []schema.MetricScope,
 	resolution int,
 ) ([]APIQuery, []schema.MetricScope, error) {
+	// Initialize both slices together
 	queries := make([]APIQuery, 0, len(metrics)*len(scopes)*len(job.Resources))
-	assignedScope := []schema.MetricScope{}
+	assignedScope := make([]schema.MetricScope, 0, len(metrics)*len(scopes)*len(job.Resources))
 
-	subcluster, scerr := archive.GetSubCluster(job.Cluster, job.SubCluster)
-	if scerr != nil {
-		return nil, nil, scerr
+	topology, err := ccms.getTopology(job.Cluster, job.SubCluster)
+	if err != nil {
+		cclog.Errorf("could not load cluster %s subCluster %s topology: %s", job.Cluster, job.SubCluster, err.Error())
+		return nil, nil, err
 	}
-	topology := subcluster.Topology
 
 	for _, metric := range metrics {
 		remoteName := metric
@@ -128,7 +129,7 @@ func (ccms *CCMetricStore) buildQueries(
 				hostQueries, hostScopes := buildScopeQueries(
 					nativeScope, requestedScope,
 					remoteName, host.Hostname,
-					&topology, hwthreads, host.Accelerators,
+					topology, hwthreads, host.Accelerators,
 					resolution,
 				)
 
@@ -163,19 +164,9 @@ func (ccms *CCMetricStore) buildNodeQueries(
 	scopes []schema.MetricScope,
 	resolution int,
 ) ([]APIQuery, []schema.MetricScope, error) {
+	// Initialize both slices together
 	queries := make([]APIQuery, 0, len(metrics)*len(scopes)*len(nodes))
-	assignedScope := []schema.MetricScope{}
-
-	// Get Topol before loop if subCluster given
-	var subClusterTopol *schema.SubCluster
-	var scterr error
-	if subCluster != "" {
-		subClusterTopol, scterr = archive.GetSubCluster(cluster, subCluster)
-		if scterr != nil {
-			cclog.Errorf("could not load cluster %s subCluster %s topology: %s", cluster, subCluster, scterr.Error())
-			return nil, nil, scterr
-		}
-	}
+	assignedScope := make([]schema.MetricScope, 0, len(metrics)*len(scopes)*len(nodes))
 
 	for _, metric := range metrics {
 		remoteName := metric
@@ -215,22 +206,22 @@ func (ccms *CCMetricStore) buildNodeQueries(
 			handledScopes = append(handledScopes, scope)
 
 			for _, hostname := range nodes {
+				var topology *schema.Topology
+				var err error
 
 				// If no subCluster given, get it by node
 				if subCluster == "" {
-					subClusterName, scnerr := archive.GetSubClusterByNode(cluster, hostname)
-					if scnerr != nil {
-						return nil, nil, scnerr
-					}
-					subClusterTopol, scterr = archive.GetSubCluster(cluster, subClusterName)
-					if scterr != nil {
-						return nil, nil, scterr
-					}
+					topology, err = ccms.getTopologyByNode(cluster, hostname)
+				} else {
+					topology, err = ccms.getTopology(cluster, subCluster)
+				}
+
+				if err != nil {
+					return nil, nil, err
 				}
 
 				// Always full node hwthread id list, no partial queries expected -> Use "topology.Node" directly where applicable
 				// Always full accelerator id list, no partial queries expected -> Use "acceleratorIds" directly where applicable
-				topology := subClusterTopol.Topology
 				acceleratorIds := topology.GetAcceleratorIDs()
 
 				// Moved check here if metric matches hardware specs
@@ -241,7 +232,7 @@ func (ccms *CCMetricStore) buildNodeQueries(
 				nodeQueries, nodeScopes := buildScopeQueries(
 					nativeScope, requestedScope,
 					remoteName, hostname,
-					&topology, topology.Node, acceleratorIds,
+					topology, topology.Node, acceleratorIds,
 					resolution,
 				)
 
@@ -278,7 +269,6 @@ func buildScopeQueries(
 	// Accelerator -> Accelerator (Use "accelerator" scope if requested scope is lower than node)
 	if nativeScope == schema.MetricScopeAccelerator && scope.LT(schema.MetricScopeNode) {
 		if scope != schema.MetricScopeAccelerator {
-			// Skip all other caught cases
 			return queries, scopes
 		}
 
@@ -451,6 +441,31 @@ func buildScopeQueries(
 		return queries, scopes
 	}
 
+	// MemoryDomain -> Socket
+	if nativeScope == schema.MetricScopeMemoryDomain && scope == schema.MetricScopeSocket {
+		memDomains, _ := topology.GetMemoryDomainsFromHWThreads(hwthreads)
+		socketToDomains, err := topology.GetMemoryDomainsBySocket(memDomains)
+		if err != nil {
+			cclog.Errorf("Error mapping memory domains to sockets, return unchanged: %v", err)
+			return queries, scopes
+		}
+
+		// Create a query for each socket
+		for _, domains := range socketToDomains {
+			queries = append(queries, APIQuery{
+				Metric:     metric,
+				Hostname:   hostname,
+				Aggregate:  true,
+				Type:       &memoryDomainString,
+				TypeIds:    intToStringSlice(domains),
+				Resolution: resolution,
+			})
+			// Add scope for each query, not just once
+			scopes = append(scopes, scope)
+		}
+		return queries, scopes
+	}
+
 	// Socket -> Socket
 	if nativeScope == schema.MetricScopeSocket && scope == schema.MetricScopeSocket {
 		sockets, _ := topology.GetSocketsFromHWThreads(hwthreads)

internal/metricstoreclient/cc-metric-store.go

@@ -71,10 +71,11 @@ import (
 // CCMetricStore is the HTTP client for communicating with cc-metric-store.
 // It manages connection details, authentication, and provides methods for querying metrics.
 type CCMetricStore struct {
-	client        http.Client // HTTP client with 10-second timeout
-	jwt           string      // JWT Bearer token for authentication
-	url           string      // Base URL of cc-metric-store instance
-	queryEndpoint string      // Full URL to query API endpoint
+	client        http.Client                 // HTTP client with 10-second timeout
+	jwt           string                      // JWT Bearer token for authentication
+	url           string                      // Base URL of cc-metric-store instance
+	queryEndpoint string                      // Full URL to query API endpoint
+	topologyCache map[string]*schema.Topology // cluster -> topology cache
 }
 
 // APIQueryRequest represents a request to the cc-metric-store query API.
@@ -133,6 +134,7 @@ func NewCCMetricStore(url string, token string) *CCMetricStore {
 		client: http.Client{
 			Timeout: 10 * time.Second,
 		},
+		topologyCache: make(map[string]*schema.Topology),
 	}
 }
 
@@ -185,6 +187,32 @@ func (ccms *CCMetricStore) doRequest(
 	return &resBody, nil
 }
 
+// getTopology returns the topology for a given cluster and subcluster, caching it if not already present
+func (ccms *CCMetricStore) getTopology(cluster, subCluster string) (*schema.Topology, error) {
+	cacheKey := fmt.Sprintf("%s:%s", cluster, subCluster)
+	if topology, ok := ccms.topologyCache[cacheKey]; ok {
+		return topology, nil
+	}
+
+	subcluster, err := archive.GetSubCluster(cluster, subCluster)
+	if err != nil {
+		return nil, err
+	}
+
+	ccms.topologyCache[cacheKey] = &subcluster.Topology
+	return &subcluster.Topology, nil
+}
+
+// getTopologyByNode returns the topology for a given cluster and node, caching it if not already present
+func (ccms *CCMetricStore) getTopologyByNode(cluster, node string) (*schema.Topology, error) {
+	subCluster, err := archive.GetSubClusterByNode(cluster, node)
+	if err != nil {
+		return nil, err
+	}
+
+	return ccms.getTopology(cluster, subCluster)
+}
+
 // LoadData retrieves time series data and statistics for the specified job and metrics.
 // It queries data for the job's time range and resources, handling scope transformations automatically.
 //
@@ -210,6 +238,12 @@ func (ccms *CCMetricStore) LoadData(
 		return nil, err
 	}
 
+	// Verify assignment is correct - log any inconsistencies for debugging
+	if len(queries) != len(assignedScope) {
+		cclog.Errorf("Critical error: queries and assignedScope have different lengths after buildQueries: %d vs %d",
+			len(queries), len(assignedScope))
+	}
+
 	req := APIQueryRequest{
 		Cluster:   job.Cluster,
 		From:      job.StartTime,
@@ -227,11 +261,37 @@ func (ccms *CCMetricStore) LoadData(
 
 	var errors []string
 	jobData := make(schema.JobData)
+
+	// Add safety check for potential index out of range errors
+	if len(resBody.Results) != len(req.Queries) || len(assignedScope) != len(req.Queries) {
+		cclog.Warnf("Mismatch in query results count: queries=%d, results=%d, assignedScope=%d",
+			len(req.Queries), len(resBody.Results), len(assignedScope))
+		if len(resBody.Results) > len(req.Queries) {
+			resBody.Results = resBody.Results[:len(req.Queries)]
+		}
+		if len(assignedScope) > len(req.Queries) {
+			assignedScope = assignedScope[:len(req.Queries)]
+		}
+	}
+
 	for i, row := range resBody.Results {
+		// Safety check to prevent index out of range errors
+		if i >= len(req.Queries) || i >= len(assignedScope) {
+			cclog.Warnf("Index out of range prevented: i=%d, queries=%d, assignedScope=%d",
+				i, len(req.Queries), len(assignedScope))
+			continue
+		}
+
 		query := req.Queries[i]
 		metric := query.Metric
 		scope := assignedScope[i]
 		mc := archive.GetMetricConfig(job.Cluster, metric)
+
+		if mc == nil {
+			cclog.Warnf("Metric config not found for %s on cluster %s", metric, job.Cluster)
+			continue
+		}
+
 		if _, ok := jobData[metric]; !ok {
 			jobData[metric] = make(map[schema.MetricScope]*schema.JobMetric)
 		}
@@ -260,8 +320,15 @@ func (ccms *CCMetricStore) LoadData(
 
 			id := (*string)(nil)
 			if query.Type != nil {
-				id = new(string)
-				*id = query.TypeIds[ndx]
+				// Check if ndx is within the bounds of TypeIds slice
+				if ndx < len(query.TypeIds) {
+					id = new(string)
+					*id = query.TypeIds[ndx]
+				} else {
+					// Log the error but continue processing
+					cclog.Warnf("TypeIds index out of range: %d with length %d for metric %s on host %s",
+						ndx, len(query.TypeIds), query.Metric, query.Hostname)
+				}
 			}
 
 			sanitizeStats(&res.Avg, &res.Min, &res.Max)
@@ -412,8 +479,15 @@ func (ccms *CCMetricStore) LoadScopedStats(
 
 			id := (*string)(nil)
 			if query.Type != nil {
-				id = new(string)
-				*id = query.TypeIds[ndx]
+				// Check if ndx is within the bounds of TypeIds slice
+				if ndx < len(query.TypeIds) {
+					id = new(string)
+					*id = query.TypeIds[ndx]
+				} else {
+					// Log the error but continue processing
+					cclog.Warnf("TypeIds index out of range: %d with length %d for metric %s on host %s",
+						ndx, len(query.TypeIds), query.Metric, query.Hostname)
+				}
 			}
 
 			sanitizeStats(&res.Avg, &res.Min, &res.Max)
@@ -561,6 +635,12 @@ func (ccms *CCMetricStore) LoadNodeListData(
 		return nil, err
 	}
 
+	// Verify assignment is correct - log any inconsistencies for debugging
+	if len(queries) != len(assignedScope) {
+		cclog.Errorf("Critical error: queries and assignedScope have different lengths after buildNodeQueries: %d vs %d",
+			len(queries), len(assignedScope))
+	}
+
 	req := APIQueryRequest{
 		Cluster:   cluster,
 		Queries:   queries,
@@ -578,17 +658,47 @@ func (ccms *CCMetricStore) LoadNodeListData(
 
 	var errors []string
 	data := make(map[string]schema.JobData)
+
+	// Add safety check for index out of range issues
+	if len(resBody.Results) != len(req.Queries) || len(assignedScope) != len(req.Queries) {
+		cclog.Warnf("Mismatch in query results count: queries=%d, results=%d, assignedScope=%d",
+			len(req.Queries), len(resBody.Results), len(assignedScope))
+		if len(resBody.Results) > len(req.Queries) {
+			resBody.Results = resBody.Results[:len(req.Queries)]
+		}
+		if len(assignedScope) > len(req.Queries) {
+			assignedScope = assignedScope[:len(req.Queries)]
+		}
+	}
+
 	for i, row := range resBody.Results {
+		// Safety check to prevent index out of range errors
+		if i >= len(req.Queries) || i >= len(assignedScope) {
+			cclog.Warnf("Index out of range prevented: i=%d, queries=%d, assignedScope=%d",
+				i, len(req.Queries), len(assignedScope))
+			continue
+		}
+
 		var query APIQuery
 		if resBody.Queries != nil {
-			query = resBody.Queries[i]
+			if i < len(resBody.Queries) {
+				query = resBody.Queries[i]
+			} else {
+				cclog.Warnf("Index out of range prevented for resBody.Queries: i=%d, len=%d",
+					i, len(resBody.Queries))
+				continue
+			}
 		} else {
 			query = req.Queries[i]
 		}
-		// qdata := res[0]
+
 		metric := query.Metric
 		scope := assignedScope[i]
 		mc := archive.GetMetricConfig(cluster, metric)
+		if mc == nil {
+			cclog.Warnf("Metric config not found for %s on cluster %s", metric, cluster)
+			continue
+		}
 
 		res := mc.Timestep
 		if len(row) > 0 {
@@ -627,8 +737,15 @@ func (ccms *CCMetricStore) LoadNodeListData(
 
 			id := (*string)(nil)
 			if query.Type != nil {
-				id = new(string)
-				*id = query.TypeIds[ndx]
+				// Check if ndx is within the bounds of TypeIds slice
+				if ndx < len(query.TypeIds) {
+					id = new(string)
+					*id = query.TypeIds[ndx]
+				} else {
+					// Log the error but continue processing
+					cclog.Warnf("TypeIds index out of range: %d with length %d for metric %s on host %s",
+						ndx, len(query.TypeIds), query.Metric, query.Hostname)
+				}
 			}
 
 			sanitizeStats(&res.Avg, &res.Min, &res.Max)

pkg/metricstore/query.go

@@ -93,6 +93,12 @@ func (ccms *InternalMetricStore) LoadData(
 		return nil, err
 	}
 
+	// Verify assignment is correct - log any inconsistencies for debugging
+	if len(queries) != len(assignedScope) {
+		cclog.Errorf("Critical error: queries and assignedScope have different lengths after buildQueries: %d vs %d",
+			len(queries), len(assignedScope))
+	}
+
 	req := APIQueryRequest{
 		Cluster:   job.Cluster,
 		From:      job.StartTime,
@@ -110,9 +116,24 @@ func (ccms *InternalMetricStore) LoadData(
 
 	var errors []string
 	jobData := make(schema.JobData)
+
+	// Add safety check for potential index out of range errors
+	if len(resBody.Results) != len(req.Queries) || len(assignedScope) != len(req.Queries) {
+		cclog.Warnf("Mismatch in query results count: queries=%d, results=%d, assignedScope=%d",
+			len(req.Queries), len(resBody.Results), len(assignedScope))
+		if len(resBody.Results) > len(req.Queries) {
+			resBody.Results = resBody.Results[:len(req.Queries)]
+		}
+		if len(assignedScope) > len(req.Queries) {
+			assignedScope = assignedScope[:len(req.Queries)]
+		}
+	}
+
 	for i, row := range resBody.Results {
-		if len(row) == 0 {
-			// No Data Found For Metric, Logged in FetchData to Warn
+		// Safety check to prevent index out of range errors
+		if i >= len(req.Queries) || i >= len(assignedScope) {
+			cclog.Warnf("Index out of range prevented: i=%d, queries=%d, assignedScope=%d",
+				i, len(req.Queries), len(assignedScope))
 			continue
 		}
 
@@ -120,6 +141,12 @@ func (ccms *InternalMetricStore) LoadData(
 		metric := query.Metric
 		scope := assignedScope[i]
 		mc := archive.GetMetricConfig(job.Cluster, metric)
+
+		if mc == nil {
+			cclog.Warnf("Metric config not found for %s on cluster %s", metric, job.Cluster)
+			continue
+		}
+
 		if _, ok := jobData[metric]; !ok {
 			jobData[metric] = make(map[schema.MetricScope]*schema.JobMetric)
 		}
@@ -148,8 +175,15 @@ func (ccms *InternalMetricStore) LoadData(
 
 			id := (*string)(nil)
 			if query.Type != nil {
-				id = new(string)
-				*id = query.TypeIds[ndx]
+				// Check if ndx is within the bounds of TypeIds slice
+				if ndx < len(query.TypeIds) {
+					id = new(string)
+					*id = query.TypeIds[ndx]
+				} else {
+					// Log the error but continue processing
+					cclog.Warnf("TypeIds index out of range: %d with length %d for metric %s on host %s",
+						ndx, len(query.TypeIds), query.Metric, query.Hostname)
+				}
 			}
 
 			sanitizeStats(&res)
@@ -650,8 +684,15 @@ func (ccms *InternalMetricStore) LoadScopedStats(
 
 			id := (*string)(nil)
 			if query.Type != nil {
-				id = new(string)
-				*id = query.TypeIds[ndx]
+				// Check if ndx is within the bounds of TypeIds slice
+				if ndx < len(query.TypeIds) {
+					id = new(string)
+					*id = query.TypeIds[ndx]
+				} else {
+					// Log the error but continue processing
+					cclog.Warnf("TypeIds index out of range: %d with length %d for metric %s on host %s",
+						ndx, len(query.TypeIds), query.Metric, query.Hostname)
+				}
 			}
 
 			sanitizeStats(&res)
@@ -823,6 +864,12 @@ func (ccms *InternalMetricStore) LoadNodeListData(
 		return nil, err
 	}
 
+	// Verify assignment is correct - log any inconsistencies for debugging
+	if len(queries) != len(assignedScope) {
+		cclog.Errorf("Critical error: queries and assignedScope have different lengths after buildNodeQueries: %d vs %d",
+			len(queries), len(assignedScope))
+	}
+
 	req := APIQueryRequest{
 		Cluster:   cluster,
 		Queries:   queries,
@@ -840,14 +887,36 @@ func (ccms *InternalMetricStore) LoadNodeListData(
 
 	var errors []string
 	data := make(map[string]schema.JobData)
+
+	// Add safety check for index out of range issues
+	if len(resBody.Results) != len(req.Queries) || len(assignedScope) != len(req.Queries) {
+		cclog.Warnf("Mismatch in query results count: queries=%d, results=%d, assignedScope=%d",
+			len(req.Queries), len(resBody.Results), len(assignedScope))
+		if len(resBody.Results) > len(req.Queries) {
+			resBody.Results = resBody.Results[:len(req.Queries)]
+		}
+		if len(assignedScope) > len(req.Queries) {
+			assignedScope = assignedScope[:len(req.Queries)]
+		}
+	}
+
 	for i, row := range resBody.Results {
-		if len(row) == 0 {
-			// No Data Found For Metric, Logged in FetchData to Warn
+		// Safety check to prevent index out of range errors
+		if i >= len(req.Queries) || i >= len(assignedScope) {
+			cclog.Warnf("Index out of range prevented: i=%d, queries=%d, assignedScope=%d",
+				i, len(req.Queries), len(assignedScope))
 			continue
 		}
+
 		var query APIQuery
 		if resBody.Queries != nil {
-			query = resBody.Queries[i]
+			if i < len(resBody.Queries) {
+				query = resBody.Queries[i]
+			} else {
+				cclog.Warnf("Index out of range prevented for resBody.Queries: i=%d, len=%d",
+					i, len(resBody.Queries))
+				continue
+			}
 		} else {
 			query = req.Queries[i]
 		}
@@ -855,6 +924,10 @@ func (ccms *InternalMetricStore) LoadNodeListData(
 		metric := query.Metric
 		scope := assignedScope[i]
 		mc := archive.GetMetricConfig(cluster, metric)
+		if mc == nil {
+			cclog.Warnf("Metric config not found for %s on cluster %s", metric, cluster)
+			continue
+		}
 
 		res := mc.Timestep
 		if len(row) > 0 {
@@ -893,8 +966,15 @@ func (ccms *InternalMetricStore) LoadNodeListData(
 
 			id := (*string)(nil)
 			if query.Type != nil {
-				id = new(string)
-				*id = query.TypeIds[ndx]
+				// Check if ndx is within the bounds of TypeIds slice
+				if ndx < len(query.TypeIds) {
+					id = new(string)
+					*id = query.TypeIds[ndx]
+				} else {
+					// Log the error but continue processing
+					cclog.Warnf("TypeIds index out of range: %d with length %d for metric %s on host %s",
+						ndx, len(query.TypeIds), query.Metric, query.Hostname)
+				}
 			}
 
 			sanitizeStats(&res)

web/frontend/src/generic/utils.js

@@ -60,6 +60,7 @@ export function init(extraInitQuery = "") {
                 topology {
                     node
                     socket
+                    memoryDomain
                     core
                     accelerators { id }
                 }
@@ -238,7 +239,7 @@ export function groupByScope(jobMetrics) {
 const scopeGranularity = {
     node: 10,
     socket: 5,
-    memorydomain: 4,
+    memoryDomain: 4,
     core: 3,
     hwthread: 2,
     accelerator: 1