Merge pull request #36322 from vespa-engine/olaa/add-snapshot-busy-metric

Add snapshot busy metric

Author: olaaun

metrics/src/main/java/ai/vespa/metrics/ConfigServerMetrics.java

@@ -71,6 +71,7 @@ public enum ConfigServerMetrics implements VespaMetrics {
     CLUSTER_LOAD_PEAK_MEMORY("cluster.load.peak.memory", Unit.FRACTION, "The peak memory load in the period considered of a certain cluster"),
     CLUSTER_LOAD_PEAK_DISK("cluster.load.peak.disk", Unit.FRACTION, "The peak disk load in the period considered of a certain cluster"),
     CLUSTER_BACKUP_AGE_FRACTION("cluster.backup.age", Unit.FRACTION, "Age of the most recent cluster backup as a fraction of the backup interval"),
+    CLUSTER_SNAPSHOT_BUSY_SECONDS("cluster.snapshot.busySeconds", Unit.SECOND, "The maximum time a snapshot has been busy (creating or restoring) for a cluster"),
 
     ZONE_WORKING("zone.working", Unit.BINARY, "The value 1 if zone is considered healthy, 0 if not. This is decided by considering the number of non-active nodes vs the number of active nodes in a zone"),
     CACHE_NODE_OBJECT_HIT_RATE("cache.nodeObject.hitRate", Unit.FRACTION, "The fraction of cache hits vs cache lookups for the node object cache"),

metrics/src/main/java/ai/vespa/metrics/set/InfrastructureMetricSet.java

@@ -71,6 +71,7 @@ private static Set<Metric> getConfigServerMetrics() {
         addMetric(metrics, ConfigServerMetrics.CLUSTER_LOAD_PEAK_MEMORY.max());
         addMetric(metrics, ConfigServerMetrics.CLUSTER_LOAD_PEAK_DISK.max());
         addMetric(metrics, ConfigServerMetrics.CLUSTER_BACKUP_AGE_FRACTION.last());
+        addMetric(metrics, ConfigServerMetrics.CLUSTER_SNAPSHOT_BUSY_SECONDS.max());
         addMetric(metrics, ConfigServerMetrics.NODES_EMPTY_EXCLUSIVE.max());
         addMetric(metrics, ConfigServerMetrics.NODES_EXPIRED_DEPROVISIONED.count());
         addMetric(metrics, ConfigServerMetrics.NODES_EXPIRED_DIRTY.count());