[ARM] Metric not found after using list metric definition

[jsarrelli]

We are consuming metrics for different resources.
The way we do it is by first listing the metrics

  def listMetricsV2(resourceId: String): Future[ResourceMetrics] =
    Source
      .fromPublisher[MetricDefinition](
        underlying
          .listByResourceAsync(resourceId)
          .timeout(Duration.ofSeconds(5))
          .retryWhen(AzureClient.retrySpec)
      )
      .runFold(List.empty[MetricDefinition])(_ :+ _)
      .map(metricDefinitions =>
        ResourceMetrics(
          resourceId,
          metricDefinitions.map { metric =>
            metric.name.value -> metric
          }.toMap,
        )
      )

and then with those metrics, we query specific ones.

....
    val datapointsF: Future[Map[Instant, Double]] = metrics
      .get(metricName)
      .map { metric =>
        val query = metric
          .defineQuery()
          .startingFrom(OffsetDateTime.ofInstant(start, ZoneId.systemDefault()))
          .endsBefore(OffsetDateTime.ofInstant(end, ZoneId.systemDefault()))
          .withAggregation(aggregation.entryName)
          .withInterval(Duration.ofSeconds(period))
      }.getOrElse {
        log.debug("Requested metric was not under Azure available ones")
        }

...

So with the metric present in the listMetric response, we are getting the following errors.

com.azure.core.management.exception.ManagementException: Status code 400, "{"code":"BadRequest","message":"Failed to find metric configuration for provider: Microsoft.Web, resource Type: serverfarms, metric: DiskQueueLength, Valid metrics: "}"

SDK Version:
"com.azure.resourcemanager" % "azure-resourcemanager" % "2.42.0"

[joshfree]

@weidongxu-microsoft could you please take a look?

[weidongxu-microsoft]

@v-hongli1 Please see if you can reproduce. The resource for the monitor metrics should be AppServicePlan.

[jsarrelli]

I think it happens for different services

com.azure.core.management.exception.ManagementException: Status code 400, "{"code":"BadRequest","message":"Failed to find metric configuration for provider: Microsoft.Sql, resource Type: servers/databases, metric: dtu_consumption_percent, Valid metrics: cpu_percent,physical_data_read_percent,connection_successful,connection_failed,connection_failed_user_error,blocked_by_firewall,availability,ledger_digest_upload_success,ledger_digest_upload_failed"}"

Here is the example on how dtu_consumption_percent exists on the listMetrics response but failed when try to query the values

[v-hongli1]

@

I think it happens for different services

com.azure.core.management.exception.ManagementException: Status code 400, "{"code":"BadRequest","message":"Failed to find metric configuration for provider: Microsoft.Sql, resource Type: servers/databases, metric: dtu_consumption_percent, Valid metrics: cpu_percent,physical_data_read_percent,connection_successful,connection_failed,connection_failed_user_error,blocked_by_firewall,availability,ledger_digest_upload_success,ledger_digest_upload_failed"}"

Here is the example on how dtu_consumption_percent exists on the listMetrics response but failed when try to query the values

The available aggregations for each metric in Microsoft.Sql/servers/databases are not exactly the same, so it is necessary to filter the metrics by the available aggregations.

[v-hongli1]

We are consuming metrics for different resources. The way we do it is by first listing the metrics

  def listMetricsV2(resourceId: String): Future[ResourceMetrics] =
    Source
      .fromPublisher[MetricDefinition](
        underlying
          .listByResourceAsync(resourceId)
          .timeout(Duration.ofSeconds(5))
          .retryWhen(AzureClient.retrySpec)
      )
      .runFold(List.empty[MetricDefinition])(_ :+ _)
      .map(metricDefinitions =>
        ResourceMetrics(
          resourceId,
          metricDefinitions.map { metric =>
            metric.name.value -> metric
          }.toMap,
        )
      )

and then with those metrics, we query specific ones.

....
    val datapointsF: Future[Map[Instant, Double]] = metrics
      .get(metricName)
      .map { metric =>
        val query = metric
          .defineQuery()
          .startingFrom(OffsetDateTime.ofInstant(start, ZoneId.systemDefault()))
          .endsBefore(OffsetDateTime.ofInstant(end, ZoneId.systemDefault()))
          .withAggregation(aggregation.entryName)
          .withInterval(Duration.ofSeconds(period))
      }.getOrElse {
        log.debug("Requested metric was not under Azure available ones")
        }

...

So with the metric present in the listMetric response, we are getting the following errors.

com.azure.core.management.exception.ManagementException: Status code 400, "{"code":"BadRequest","message":"Failed to find metric configuration for provider: Microsoft.Web, resource Type: serverfarms, metric: DiskQueueLength, Valid metrics: "}"

SDK Version: "com.azure.resourcemanager" % "azure-resourcemanager" % "2.42.0"

1. The aggregation of Microsoft.Web/serverfarms only supports "None", "Average", "Minimum", "Maximum", "Total", "Count"

2. The interval only supports PT1M, PT5M, PT15M, PT30M, PT1H, PT6H, PT12H, and P1D.

Please confirm that the values ​​of Aggregation and Interval in the above code are correct.

[jsarrelli]

In both cases, I'm aggregating by Average using a PT1M time grain