Merge pull request #513 from ClusterCockpit/dev

Dev

ReleaseNotes.md

@@ -271,6 +271,8 @@ For release specific notes visit the [ClusterCockpit Documentation](https://clus
 
 ## Known issues
 
+- The new dynamic memory management is not bullet proof yet across restarts. We
+  will fix that in a subsequent patch release
 - Currently energy footprint metrics of type energy are ignored for calculating
   total energy.
 - With energy footprint metrics of type power the unit is ignored and it is

cmd/cc-backend/main.go

@@ -395,6 +395,7 @@ func runServer(ctx context.Context) error {
 
 	// Set GC percent if not configured
 	if os.Getenv(envGOGC) == "" {
+		// trigger GC when heap grows 15% above the previous live set
 		debug.SetGCPercent(15)
 	}
 	runtime.SystemdNotify(true, "running")

web/frontend/src/generic/joblist/JobListRow.svelte

@@ -234,8 +234,6 @@
             cluster={clusterInfos.find((c) => c.name == job.cluster)}
             subCluster={job.subCluster}
             isShared={job.shared != "none"}
-            numhwthreads={job.numHWThreads}
-            numaccs={job.numAcc}
             zoomState={zoomStates[metric.data.name] || null}
             thresholdState={thresholdStates[metric.data.name] || null}
             {plotSync}

web/frontend/src/generic/plots/MetricPlot.svelte

@@ -43,8 +43,6 @@
     subCluster,
     isShared = false,
     forNode = false,
-    numhwthreads = 0,
-    numaccs = 0,
     zoomState = null,
     thresholdState = null,
     extendedLegendData = null,
@@ -83,9 +81,7 @@
   const thresholds = $derived(findJobAggregationThresholds(
     subClusterTopology,
     metricConfig,
-    scope,
+    scope
-    numhwthreads,
-    numaccs
   ));
   const longestSeries = $derived.by(() => {
     if (useStatsSeries) {
@@ -276,9 +272,7 @@
   function findJobAggregationThresholds(
     subClusterTopology,
     metricConfig,
-    scope,
+    scope
-    numhwthreads,
-    numaccs
   ) {
 
     if (!subClusterTopology || !metricConfig || !scope) {
@@ -303,22 +297,14 @@
     }
 
     if (metricConfig?.aggregation == "sum") {
-      // Scale Thresholds
-      let fraction;
-      if (numaccs > 0) fraction = subClusterTopology.accelerators.length / numaccs;
-      else if (numhwthreads > 0) fraction = subClusterTopology.core.length / numhwthreads;
-      else fraction = 1; // Fallback
-
       let divisor;
-      // Exclusive: Fraction = 1; Shared: Fraction > 1
+      if (scope == 'node')              divisor = 1 // Node Scope: Always return unscaled (Maximum Scope)
-      if (scope == 'node')              divisor = fraction;
+      // Partial Scopes: Get from Topologies
-      // Cap divisor at number of available sockets or domains
+      else if (scope == 'socket')       divisor = subClusterTopology?.socket?.length || 1;
-      else if (scope == 'socket')       divisor = (fraction < subClusterTopology.socket.length) ? subClusterTopology.socket.length : fraction;
+      else if (scope == "memoryDomain") divisor = subClusterTopology?.memoryDomain?.length || 1;
-      else if (scope == "memoryDomain") divisor = (fraction < subClusterTopology.memoryDomain.length) ? subClusterTopology.socket.length : fraction;
+      else if (scope == "core")         divisor = subClusterTopology?.core?.length || 1;
-      // Use Maximum Division for Smallest Scopes
+      else if (scope == "hwthread")     divisor = subClusterTopology?.node?.length || 1;
-      else if (scope == "core")         divisor = subClusterTopology.core.length;
+      else if (scope == "accelerator")  divisor = subClusterTopology?.accelerators?.length || 1;
-      else if (scope == "hwthread")     divisor = subClusterTopology.core.length; // alt. name for core
-      else if (scope == "accelerator")  divisor = subClusterTopology.accelerators.length;
       else {
         console.log('Unknown scope, return default aggregation thresholds for sum', scope)
         divisor = 1;

web/frontend/src/job/Metric.svelte

@@ -178,8 +178,6 @@
       timestep={selectedData.timestep}
       scope={selectedScope}
       metric={metricName}
-      numaccs={job.numAcc}
-      numhwthreads={job.numHWThreads}
       series={selectedSeries}
       {isShared}
       {zoomState}
@@ -194,8 +192,6 @@
       timestep={selectedData.timestep}
       scope={selectedScope}
       metric={metricName}
-      numaccs={job.numAcc}
-      numhwthreads={job.numHWThreads}
       series={selectedSeries}
       {isShared}
       {zoomState}