mxm commented on code in PR #617:
URL:
https://github.com/apache/flink-kubernetes-operator/pull/617#discussion_r1224078049
##########
flink-kubernetes-operator-autoscaler/src/main/java/org/apache/flink/kubernetes/operator/autoscaler/ScalingMetricCollector.java:
##########
@@ -101,6 +100,7 @@ public CollectedMetricHistory updateMetrics(
metricHistory.clear();
metricCollectionStartTs = now;
}
+ var topology = getJobTopology(flinkService, cr, conf, autoscalerInfo);
Review Comment:
Any reason for moving this?
##########
flink-kubernetes-operator-autoscaler/src/main/java/org/apache/flink/kubernetes/operator/autoscaler/JobAutoScalerImpl.java:
##########
@@ -102,10 +108,8 @@ public boolean scale(FlinkResourceContext<? extends
AbstractFlinkResource<?, ?>>
resource, autoScalerInfo, ctx.getFlinkService(),
conf);
LOG.debug("Evaluating scaling metrics for {}", collectedMetrics);
- var evaluatedMetrics = evaluator.evaluate(conf, collectedMetrics);
+ evaluatedMetrics = evaluator.evaluate(conf, collectedMetrics);
LOG.debug("Scaling metrics evaluated: {}", evaluatedMetrics);
- lastEvaluatedMetrics.put(resourceId, evaluatedMetrics);
- flinkMetrics.registerScalingMetrics(() ->
lastEvaluatedMetrics.get(resourceId));
if (!collectedMetrics.isFullyCollected()) {
// We have done an upfront evaluation, but we are not ready
for scaling.
Review Comment:
I think it makes sense to only set the recommended parallelism metric when
we are actually recommending a parallelism. In all other cases, we should clear
the recommended parallelism metric.
##########
flink-kubernetes-operator-autoscaler/src/test/java/org/apache/flink/kubernetes/operator/autoscaler/RecommendedParallelismTest.java:
##########
@@ -0,0 +1,296 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.flink.kubernetes.operator.autoscaler;
+
+import org.apache.flink.api.common.JobID;
+import org.apache.flink.api.common.JobStatus;
+import org.apache.flink.configuration.Configuration;
+import org.apache.flink.kubernetes.operator.OperatorTestBase;
+import org.apache.flink.kubernetes.operator.TestUtils;
+import org.apache.flink.kubernetes.operator.api.FlinkDeployment;
+import
org.apache.flink.kubernetes.operator.autoscaler.config.AutoScalerOptions;
+import org.apache.flink.kubernetes.operator.autoscaler.metrics.FlinkMetric;
+import org.apache.flink.kubernetes.operator.autoscaler.metrics.ScalingMetric;
+import org.apache.flink.kubernetes.operator.autoscaler.topology.JobTopology;
+import org.apache.flink.kubernetes.operator.autoscaler.topology.VertexInfo;
+import org.apache.flink.kubernetes.operator.config.FlinkConfigManager;
+import org.apache.flink.kubernetes.operator.reconciler.ReconciliationUtils;
+import org.apache.flink.kubernetes.operator.utils.EventCollector;
+import org.apache.flink.kubernetes.operator.utils.EventRecorder;
+import org.apache.flink.runtime.jobgraph.JobVertexID;
+import org.apache.flink.runtime.rest.messages.job.metrics.AggregatedMetric;
+
+import io.fabric8.kubernetes.api.model.HasMetadata;
+import io.fabric8.kubernetes.client.KubernetesClient;
+import io.fabric8.kubernetes.client.server.mock.EnableKubernetesMockClient;
+import io.javaoperatorsdk.operator.processing.event.ResourceID;
+import lombok.Getter;
+import org.jetbrains.annotations.NotNull;
+import org.junit.jupiter.api.BeforeEach;
+import org.junit.jupiter.api.Test;
+
+import java.time.Clock;
+import java.time.Duration;
+import java.time.Instant;
+import java.time.ZoneId;
+import java.util.Map;
+import java.util.Set;
+import java.util.stream.Collectors;
+
+import static
org.apache.flink.kubernetes.operator.autoscaler.metrics.ScalingMetric.PARALLELISM;
+import static
org.apache.flink.kubernetes.operator.autoscaler.metrics.ScalingMetric.RECOMMENDED_PARALLELISM;
+import static org.junit.jupiter.api.Assertions.assertEquals;
+import static org.junit.jupiter.api.Assertions.assertNull;
+
+/** Test for recommended parallelism. */
+@EnableKubernetesMockClient(crud = true)
+public class RecommendedParallelismTest extends OperatorTestBase {
Review Comment:
I agree with both of you that (a) having a dedicated test is preferable (b)
duplicating most of the logic of an already hard to maintain test should be
avoided. I think a compromise would be to add assertions to the existing end to
end test and add a dedicated unit/integration test which tests only the core
assumptions without going through the entire lifecycle.
Adding a test method to ScalingExecutor should be sufficient IMHO to test
that the recommended parallelism is added. The cleanup can be tested in the end
to end test. If we want to go all the way, we could test the cleanup in
JobAutoscalerImpl without the entire end to end setup, e.g. look at
JobAutoscalerImplTest#testErrorReporting which only tests the error reporting.
Since you are cleaning up in the finally block, the test setup should work
similarly.
##########
flink-kubernetes-operator-autoscaler/src/test/java/org/apache/flink/kubernetes/operator/autoscaler/RecommendedParallelismTest.java:
##########
@@ -0,0 +1,296 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.flink.kubernetes.operator.autoscaler;
+
+import org.apache.flink.api.common.JobID;
+import org.apache.flink.api.common.JobStatus;
+import org.apache.flink.configuration.Configuration;
+import org.apache.flink.kubernetes.operator.OperatorTestBase;
+import org.apache.flink.kubernetes.operator.TestUtils;
+import org.apache.flink.kubernetes.operator.api.FlinkDeployment;
+import
org.apache.flink.kubernetes.operator.autoscaler.config.AutoScalerOptions;
+import org.apache.flink.kubernetes.operator.autoscaler.metrics.FlinkMetric;
+import org.apache.flink.kubernetes.operator.autoscaler.metrics.ScalingMetric;
+import org.apache.flink.kubernetes.operator.autoscaler.topology.JobTopology;
+import org.apache.flink.kubernetes.operator.autoscaler.topology.VertexInfo;
+import org.apache.flink.kubernetes.operator.config.FlinkConfigManager;
+import org.apache.flink.kubernetes.operator.reconciler.ReconciliationUtils;
+import org.apache.flink.kubernetes.operator.utils.EventCollector;
+import org.apache.flink.kubernetes.operator.utils.EventRecorder;
+import org.apache.flink.runtime.jobgraph.JobVertexID;
+import org.apache.flink.runtime.rest.messages.job.metrics.AggregatedMetric;
+
+import io.fabric8.kubernetes.api.model.HasMetadata;
+import io.fabric8.kubernetes.client.KubernetesClient;
+import io.fabric8.kubernetes.client.server.mock.EnableKubernetesMockClient;
+import io.javaoperatorsdk.operator.processing.event.ResourceID;
+import lombok.Getter;
+import org.jetbrains.annotations.NotNull;
+import org.junit.jupiter.api.BeforeEach;
+import org.junit.jupiter.api.Test;
+
+import java.time.Clock;
+import java.time.Duration;
+import java.time.Instant;
+import java.time.ZoneId;
+import java.util.Map;
+import java.util.Set;
+import java.util.stream.Collectors;
+
+import static
org.apache.flink.kubernetes.operator.autoscaler.metrics.ScalingMetric.PARALLELISM;
+import static
org.apache.flink.kubernetes.operator.autoscaler.metrics.ScalingMetric.RECOMMENDED_PARALLELISM;
+import static org.junit.jupiter.api.Assertions.assertEquals;
+import static org.junit.jupiter.api.Assertions.assertNull;
+
+/** Test for recommended parallelism. */
+@EnableKubernetesMockClient(crud = true)
+public class RecommendedParallelismTest extends OperatorTestBase {
Review Comment:
I agree with both of you that (a) having a dedicated test is preferable (b)
duplicating most of the logic of an already hard to maintain test should be
avoided. I think a compromise would be to add assertions to the existing end to
end test and add a dedicated unit/integration test which tests only the core
assumptions without going through the entire lifecycle.
##########
flink-kubernetes-operator-autoscaler/src/main/java/org/apache/flink/kubernetes/operator/autoscaler/ScalingMetricEvaluator.java:
##########
@@ -137,12 +138,13 @@ private Map<ScalingMetric, EvaluatedScalingMetric>
evaluateMetrics(
latestVertexMetrics.get(TRUE_PROCESSING_RATE),
getAverage(TRUE_PROCESSING_RATE, vertex,
metricsHistory)));
- evaluatedMetrics.put(
- PARALLELISM,
EvaluatedScalingMetric.of(topology.getParallelisms().get(vertex)));
+ var parallelism =
EvaluatedScalingMetric.of(topology.getParallelisms().get(vertex));
+ evaluatedMetrics.put(PARALLELISM, parallelism);
+ evaluatedMetrics.put(RECOMMENDED_PARALLELISM, parallelism);
Review Comment:
I feel like we can't add it here yet because we don't know yet what the
recommended parallelism is.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
