Allow custom metrics for the metrics monitor health check

* Update METRICS monitor to configure custom metrics for running and queued queries

Author: ricardocardante

docs/installation.md

@@ -394,6 +394,24 @@ It retrieves the number of running and queued queries for use with
 the `QueryCountBasedRouter` (either `METRICS` or `JDBC` must be enabled if
 `QueryCountBasedRouter` is used).
 
+By default, it uses the `trino_execution_name_QueryManager_RunningQueries` and
+`trino_execution_name_QueryManager_QueuedQueries` to track the number of running
+and queued queries respectively, however these metrics can be configured as follows:
+
+```yaml
+monitor:
+    runningQueriesMetricName: io_starburst_galaxy_name_GalaxyMetrics_RunningQueries
+    queuedQueriesMetricName: io_starburst_galaxy_name_GalaxyMetrics_QueuedQueries
+```
+
+Similarly, by default the monitor pulls the metrics using the `/metrics` endpoint, but it
+can be updated to use another one:
+
+```yaml
+monitor:
+    metricsEndpoint: /v1/metrics
+```
+
 This monitor allows customizing health definitions by comparing metrics to fixed
 values. This is configured through two maps: `metricMinimumValues` and 
 `metricMaximumValues`. The keys of these maps are the metric names, and the values

gateway-ha/src/main/java/io/trino/gateway/ha/clustermonitor/ClusterStatsMetricsMonitor.java

@@ -46,15 +46,15 @@
 public class ClusterStatsMetricsMonitor
         implements ClusterStatsMonitor
 {
-    public static final String RUNNING_QUERIES_METRIC = "trino_execution_name_QueryManager_RunningQueries";
-    public static final String QUEUED_QUERIES_METRIC = "trino_execution_name_QueryManager_QueuedQueries";
     private static final Logger log = Logger.get(ClusterStatsMetricsMonitor.class);
 
     private final HttpClient httpClient;
     private final int retries;
     private final MetricsResponseHandler metricsResponseHandler;
     private final Header identityHeader;
     private final String metricsEndpoint;
+    private final String runningQueriesMetricName;
+    private final String queuedQueriesMetricName;
     private final ImmutableSet<String> metricNames;
     private final Map<String, Float> metricMinimumValues;
     private final Map<String, Float> metricMaximumValues;
@@ -71,10 +71,12 @@ public ClusterStatsMetricsMonitor(HttpClient httpClient, BackendStateConfigurati
             identityHeader = new Header("X-Trino-User", backendStateConfiguration.getUsername());
         }
         metricsEndpoint = monitorConfiguration.getMetricsEndpoint();
+        runningQueriesMetricName = monitorConfiguration.getRunningQueriesMetricName();
+        queuedQueriesMetricName = monitorConfiguration.getQueuedQueriesMetricName();
         metricMinimumValues = ImmutableMap.copyOf(monitorConfiguration.getMetricMinimumValues());
         metricMaximumValues = ImmutableMap.copyOf(monitorConfiguration.getMetricMaximumValues());
         metricNames = ImmutableSet.<String>builder()
-                .add(RUNNING_QUERIES_METRIC, QUEUED_QUERIES_METRIC)
+                .add(runningQueriesMetricName, queuedQueriesMetricName)
                 .addAll(metricMinimumValues.keySet())
                 .addAll(metricMaximumValues.keySet())
                 .build();
@@ -117,8 +119,8 @@ public ClusterStats monitor(ProxyBackendConfiguration backend)
         }
         return ClusterStats.builder(backend.getName())
                 .trinoStatus(TrinoStatus.HEALTHY)
-                .runningQueryCount((int) Float.parseFloat(metrics.get(RUNNING_QUERIES_METRIC)))
-                .queuedQueryCount((int) Float.parseFloat(metrics.get(QUEUED_QUERIES_METRIC)))
+                .runningQueryCount((int) Float.parseFloat(metrics.get(runningQueriesMetricName)))
+                .queuedQueryCount((int) Float.parseFloat(metrics.get(queuedQueriesMetricName)))
                 .proxyTo(backend.getProxyTo())
                 .externalUrl(backend.getExternalUrl())
                 .routingGroup(backend.getRoutingGroup())

gateway-ha/src/main/java/io/trino/gateway/ha/config/MonitorConfiguration.java

@@ -33,6 +33,10 @@ public class MonitorConfiguration
 
     private String metricsEndpoint = "/metrics";
 
+    private String runningQueriesMetricName = "trino_execution_name_QueryManager_RunningQueries";
+
+    private String queuedQueriesMetricName = "trino_execution_name_QueryManager_QueuedQueries";
+
     // Require 1 node for health by default. This configuration only applies to the ClusterStatsMetricsMonitor
     private Map<String, Float> metricMinimumValues = ImmutableMap.of("trino_metadata_name_DiscoveryNodeManager_ActiveNodeCount", 1f);
 
@@ -90,6 +94,26 @@ public void setMetricsEndpoint(String metricsEndpoint)
         this.metricsEndpoint = metricsEndpoint;
     }
 
+    public String getRunningQueriesMetricName()
+    {
+        return runningQueriesMetricName;
+    }
+
+    public void setRunningQueriesMetricName(String runningQueriesMetricName)
+    {
+        this.runningQueriesMetricName = runningQueriesMetricName;
+    }
+
+    public String getQueuedQueriesMetricName()
+    {
+        return queuedQueriesMetricName;
+    }
+
+    public void setQueuedQueriesMetricName(String queuedQueriesMetricName)
+    {
+        this.queuedQueriesMetricName = queuedQueriesMetricName;
+    }
+
     public Map<String, Float> getMetricMinimumValues()
     {
         return metricMinimumValues;