use success metric instead of total metric

Signed-off-by: Rahul Karajgikar <[email protected]>
opensearch-project · Sep 23, 2024 · 48e9fa2 · 48e9fa2
1 parent 065ba11
commit 48e9fa2
Show file tree

Hide file tree

Showing 3 changed files with 58 additions and 60 deletions.
diff --git a/server/src/main/java/org/opensearch/cluster/ClusterManagerMetrics.java b/server/src/main/java/org/opensearch/cluster/ClusterManagerMetrics.java
@@ -35,7 +35,7 @@ public final class ClusterManagerMetrics {
     public final Counter leaderCheckFailureCounter;
     public final Counter followerChecksFailureCounter;
     public final Counter asyncFetchFailureCounter;
-    public final Counter asyncFetchTotalCounter;
+    public final Counter asyncFetchSuccessCounter;
 
     public ClusterManagerMetrics(MetricsRegistry metricsRegistry) {
         clusterStateAppliersHistogram = metricsRegistry.createHistogram(
@@ -78,8 +78,8 @@ public ClusterManagerMetrics(MetricsRegistry metricsRegistry) {
             "Counter for number of failed async fetches",
             COUNTER_METRICS_UNIT
         );
-        asyncFetchTotalCounter = metricsRegistry.createCounter(
-            "async.fetch.total.count",
+        asyncFetchSuccessCounter = metricsRegistry.createCounter(
+            "async.fetch.success.count",
             "Counter for total number of async fetches",
             COUNTER_METRICS_UNIT
         );

diff --git a/server/src/main/java/org/opensearch/gateway/AsyncShardFetchCache.java b/server/src/main/java/org/opensearch/gateway/AsyncShardFetchCache.java
@@ -167,7 +167,7 @@ Map<DiscoveryNode, K> getCacheData(DiscoveryNodes nodes, Set<String> failedNodes
     void processResponses(List<K> responses, long fetchingRound) {
         for (K response : responses) {
             BaseNodeEntry nodeEntry = getCache().get(response.getNode().getId());
-            clusterManagerMetrics.incrementCounter(clusterManagerMetrics.asyncFetchTotalCounter, 1.0);
+            clusterManagerMetrics.incrementCounter(clusterManagerMetrics.asyncFetchSuccessCounter, 1.0);
             if (nodeEntry != null) {
                 if (validateNodeResponse(nodeEntry, fetchingRound)) {
                     // if the entry is there, for the right fetching round and not marked as failed already, process it
@@ -197,8 +197,6 @@ private boolean validateNodeResponse(BaseNodeEntry nodeEntry, long fetchingRound
     }
 
     private void handleNodeFailure(BaseNodeEntry nodeEntry, FailedNodeException failure, long fetchingRound) {
-        clusterManagerMetrics.incrementCounter(clusterManagerMetrics.asyncFetchTotalCounter, 1.0);
-        clusterManagerMetrics.incrementCounter(clusterManagerMetrics.asyncFetchFailureCounter, 1.0);
         if (nodeEntry.getFetchingRound() != fetchingRound) {
             assert nodeEntry.getFetchingRound() > fetchingRound : "node entries only replaced by newer rounds";
             logger.trace(
@@ -230,6 +228,7 @@ boolean retryableException(Throwable unwrappedCause) {
     void processFailures(List<FailedNodeException> failures, long fetchingRound) {
         for (FailedNodeException failure : failures) {
             logger.trace("processing failure {} for [{}]", failure, type);
+            clusterManagerMetrics.incrementCounter(clusterManagerMetrics.asyncFetchFailureCounter, 1.0);
             BaseNodeEntry nodeEntry = getCache().get(failure.nodeId());
             if (nodeEntry != null) {
                 handleNodeFailure(nodeEntry, failure, fetchingRound);