[
https://issues.apache.org/jira/browse/HIVE-26277?focusedWorklogId=808315&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-808315
]
ASF GitHub Bot logged work on HIVE-26277:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 13/Sep/22 14:19
Start Date: 13/Sep/22 14:19
Worklog Time Spent: 10m
Work Description: asolimando commented on code in PR #3339:
URL: https://github.com/apache/hive/pull/3339#discussion_r969692288
##########
standalone-metastore/metastore-server/src/test/java/org/apache/hadoop/hive/metastore/columnstats/aggr/DateColumnStatsAggregatorTest.java:
##########
@@ -0,0 +1,279 @@
+/*
+ * 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.hadoop.hive.metastore.columnstats.aggr;
+
+import org.apache.hadoop.hive.metastore.TableType;
+import org.apache.hadoop.hive.metastore.annotation.MetastoreUnitTest;
+import org.apache.hadoop.hive.metastore.api.ColumnStatisticsData;
+import org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj;
+import org.apache.hadoop.hive.metastore.api.Date;
+import org.apache.hadoop.hive.metastore.api.FieldSchema;
+import org.apache.hadoop.hive.metastore.api.MetaException;
+import org.apache.hadoop.hive.metastore.api.Table;
+import org.apache.hadoop.hive.metastore.columnstats.ColStatsBuilder;
+import
org.apache.hadoop.hive.metastore.utils.MetaStoreServerUtils.ColStatsObjWithSourceInfo;
+import org.junit.Assert;
+import org.junit.Test;
+import org.junit.experimental.categories.Category;
+
+import java.util.Arrays;
+import java.util.Collections;
+import java.util.List;
+
+import static
org.apache.hadoop.hive.metastore.StatisticsTestUtils.createStatsWithInfo;
+
+@Category(MetastoreUnitTest.class)
+public class DateColumnStatsAggregatorTest {
+
+ private static final Table TABLE = new Table("dummy", "db", "hive", 0, 0,
+ 0, null, null, Collections.emptyMap(), null, null,
+ TableType.MANAGED_TABLE.toString());
+ private static final FieldSchema COL = new FieldSchema("col", "int", "");
+
+ private static final Date DATE_1 = new Date(1);
+ private static final Date DATE_2 = new Date(2);
+ private static final Date DATE_3 = new Date(3);
+ private static final Date DATE_4 = new Date(4);
+ private static final Date DATE_5 = new Date(5);
+ private static final Date DATE_6 = new Date(6);
+ private static final Date DATE_7 = new Date(7);
+ private static final Date DATE_8 = new Date(8);
+ private static final Date DATE_9 = new Date(9);
+
+ @Test
+ public void testAggregateSingleStat() throws MetaException {
+ List<String> partitions = Collections.singletonList("part1");
+
+ ColumnStatisticsData data1 = new
ColStatsBuilder<>(Date.class).numNulls(1).numDVs(2).low(DATE_1).high(DATE_4)
+ .hll(DATE_1.getDaysSinceEpoch(), DATE_4.getDaysSinceEpoch()).build();
+ List<ColStatsObjWithSourceInfo> statsList =
+ Collections.singletonList(createStatsWithInfo(data1, TABLE, COL,
partitions.get(0)));
+
+ DateColumnStatsAggregator aggregator = new DateColumnStatsAggregator();
+ ColumnStatisticsObj computedStatsObj = aggregator.aggregate(statsList,
partitions, true);
+
+ Assert.assertEquals(data1, computedStatsObj.getStatsData());
+ }
+
+ @Test
+ public void testAggregateSingleStatWhenNullValues() throws MetaException {
+ List<String> partitions = Collections.singletonList("part1");
+
+ ColumnStatisticsData data1 = new
ColStatsBuilder<>(Date.class).numNulls(1).numDVs(2).build();
+ List<ColStatsObjWithSourceInfo> statsList =
+ Collections.singletonList(createStatsWithInfo(data1, TABLE, COL,
partitions.get(0)));
+
+ DateColumnStatsAggregator aggregator = new DateColumnStatsAggregator();
+ ColumnStatisticsObj computedStatsObj = aggregator.aggregate(statsList,
partitions, true);
+ Assert.assertEquals(data1, computedStatsObj.getStatsData());
+
+ aggregator.useDensityFunctionForNDVEstimation = true;
+ computedStatsObj = aggregator.aggregate(statsList, partitions, true);
+ Assert.assertEquals(data1, computedStatsObj.getStatsData());
+
+ aggregator.useDensityFunctionForNDVEstimation = false;
+ aggregator.ndvTuner = 1;
+ // ndv tuner does not have any effect because min numDVs and max numDVs
coincide (we have a single stats)
+ computedStatsObj = aggregator.aggregate(statsList, partitions, true);
+ Assert.assertEquals(data1, computedStatsObj.getStatsData());
+ }
+
+ @Test
+ public void testAggregateMultipleStatsWhenSomeNullValues() throws
MetaException {
+ List<String> partitions = Arrays.asList("part1", "part2");
+
+ long[] values1 = { DATE_1.getDaysSinceEpoch(), DATE_2.getDaysSinceEpoch()
};
+ ColumnStatisticsData data1 = new
ColStatsBuilder<>(Date.class).numNulls(1).numDVs(2)
+ .low(DATE_1).high(DATE_2).hll(values1).build();
+ ColumnStatisticsData data2 = new
ColStatsBuilder<>(Date.class).numNulls(2).numDVs(3).build();
+
+ List<ColStatsObjWithSourceInfo> statsList =
Arrays.asList(createStatsWithInfo(data1, TABLE, COL, partitions.get(0)),
+ createStatsWithInfo(data2, TABLE, COL, partitions.get(1)));
+
+ DateColumnStatsAggregator aggregator = new DateColumnStatsAggregator();
+
+ ColumnStatisticsObj computedStatsObj = aggregator.aggregate(statsList,
partitions, true);
+ ColumnStatisticsData expectedStats = new
ColStatsBuilder<>(Date.class).numNulls(3).numDVs(3)
+ .low(DATE_1).high(DATE_2).hll(values1).build();
+ Assert.assertEquals(expectedStats, computedStatsObj.getStatsData());
+
+ aggregator.useDensityFunctionForNDVEstimation = true;
+ computedStatsObj = aggregator.aggregate(statsList, partitions, true);
+ expectedStats = new ColStatsBuilder<>(Date.class).numNulls(3).numDVs(4)
+ .low(DATE_1).high(DATE_2).hll(values1).build();
+ Assert.assertEquals(expectedStats, computedStatsObj.getStatsData());
+
+ aggregator.useDensityFunctionForNDVEstimation = false;
+ aggregator.ndvTuner = 1;
+ computedStatsObj = aggregator.aggregate(statsList, partitions, true);
+ expectedStats = new ColStatsBuilder<>(Date.class).numNulls(3).numDVs(5)
+ .low(DATE_1).high(DATE_2).hll(values1).build();
+ Assert.assertEquals(expectedStats, computedStatsObj.getStatsData());
+ }
+
+ @Test
+ public void testAggregateMultiStatsWhenAllAvailable() throws MetaException {
+ List<String> partitions = Arrays.asList("part1", "part2", "part3");
+
+ long[] values1 = { DATE_1.getDaysSinceEpoch(), DATE_2.getDaysSinceEpoch(),
DATE_3.getDaysSinceEpoch() };
+ ColumnStatisticsData data1 = new
ColStatsBuilder<>(Date.class).numNulls(1).numDVs(3)
+ .low(DATE_1).high(DATE_3).hll(values1).build();
+
+ long[] values2 = { DATE_3.getDaysSinceEpoch(), DATE_4.getDaysSinceEpoch(),
DATE_5.getDaysSinceEpoch() };
+ ColumnStatisticsData data2 = new
ColStatsBuilder<>(Date.class).numNulls(2).numDVs(3)
+ .low(DATE_3).high(DATE_5).hll(values2).build();
+
+ long[] values3 = { DATE_6.getDaysSinceEpoch(), DATE_7.getDaysSinceEpoch()
};
+ ColumnStatisticsData data3 = new
ColStatsBuilder<>(Date.class).numNulls(3).numDVs(2)
+ .low(DATE_6).high(DATE_7).hll(values3).build();
+
+ List<ColStatsObjWithSourceInfo> statsList =
Arrays.asList(createStatsWithInfo(data1, TABLE, COL, partitions.get(0)),
+ createStatsWithInfo(data2, TABLE, COL, partitions.get(1)),
createStatsWithInfo(data3, TABLE, COL, partitions.get(2)));
+
+ DateColumnStatsAggregator aggregator = new DateColumnStatsAggregator();
+ ColumnStatisticsObj computedStatsObj = aggregator.aggregate(statsList,
partitions, true);
+
+ // the aggregation does not update hll, only numNDVs is, it keeps the
first hll
+ ColumnStatisticsData expectedStats = new
ColStatsBuilder<>(Date.class).numNulls(6).numDVs(7)
+ .low(DATE_1).high(DATE_7).hll(values1).build();
+
+ Assert.assertEquals(expectedStats, computedStatsObj.getStatsData());
+ }
+
+ @Test
+ public void testAggregateMultiStatsWhenUnmergeableBitVectors() throws
MetaException {
+ List<String> partitions = Arrays.asList("part1", "part2", "part3");
+
+ long[] values1 = { DATE_1.getDaysSinceEpoch(), DATE_2.getDaysSinceEpoch(),
DATE_3.getDaysSinceEpoch() };
+ ColumnStatisticsData data1 = new
ColStatsBuilder<>(Date.class).numNulls(1).numDVs(3)
+ .low(DATE_1).high(DATE_3).fmSketch(values1).build();
+ long[] values2 = { DATE_3.getDaysSinceEpoch(), DATE_4.getDaysSinceEpoch(),
DATE_5.getDaysSinceEpoch() };
+ ColumnStatisticsData data2 = new
ColStatsBuilder<>(Date.class).numNulls(2).numDVs(3)
+ .low(DATE_3).high(DATE_5).hll(values2).build();
+ long[] values3 = { DATE_1.getDaysSinceEpoch(), DATE_2.getDaysSinceEpoch(),
DATE_6.getDaysSinceEpoch(),
+ DATE_8.getDaysSinceEpoch() };
+ ColumnStatisticsData data3 = new
ColStatsBuilder<>(Date.class).numNulls(3).numDVs(4)
+ .low(DATE_1).high(DATE_8).hll(values3).build();
+
+ List<ColStatsObjWithSourceInfo> statsList =
Arrays.asList(createStatsWithInfo(data1, TABLE, COL, partitions.get(0)),
+ createStatsWithInfo(data2, TABLE, COL, partitions.get(1)),
createStatsWithInfo(data3, TABLE, COL, partitions.get(2)));
+
+ DateColumnStatsAggregator aggregator = new DateColumnStatsAggregator();
+
+ ColumnStatisticsObj computedStatsObj = aggregator.aggregate(statsList,
partitions, true);
+ // the aggregation does not update the bitvector, only numDVs is, it keeps
the first bitvector;
+ // numDVs is set to the maximum among all stats when non-mergeable
bitvectors are detected
+ ColumnStatisticsData expectedStats = new
ColStatsBuilder<>(Date.class).numNulls(6).numDVs(4)
+ .low(DATE_1).high(DATE_8).fmSketch(values1).build();
+ Assert.assertEquals(expectedStats, computedStatsObj.getStatsData());
+
+ aggregator.useDensityFunctionForNDVEstimation = true;
+ computedStatsObj = aggregator.aggregate(statsList, partitions, true);
+ // the use of the density function leads to a different estimation for
numNDV
+ expectedStats = new ColStatsBuilder<>(Date.class).numNulls(6).numDVs(6)
+ .low(DATE_1).high(DATE_8).fmSketch(values1).build();
+ Assert.assertEquals(expectedStats, computedStatsObj.getStatsData());
+
+ // here the ndv lower bound is 4 (the highest individual numDVs), the
higher bound is 10 (3 + 3 + 4, that is the
+ // sum of all the numDVs for all partitions), ndv tuner influences the
choice between the lower bound
+ // (ndvTuner = 0) and the higher bound (ndvTuner = 1), and intermediate
values for ndvTuner in the range (0, 1)
+ aggregator.useDensityFunctionForNDVEstimation = false;
+
+ aggregator.ndvTuner = 0;
+ computedStatsObj = aggregator.aggregate(statsList, partitions, true);
+ expectedStats = new ColStatsBuilder<>(Date.class).numNulls(6).numDVs(4)
+ .low(DATE_1).high(DATE_8).fmSketch(values1).build();
+ Assert.assertEquals(expectedStats, computedStatsObj.getStatsData());
+
+ aggregator.ndvTuner = 0.5;
+ computedStatsObj = aggregator.aggregate(statsList, partitions, true);
+ expectedStats = new ColStatsBuilder<>(Date.class).numNulls(6).numDVs(7)
+ .low(DATE_1).high(DATE_8).fmSketch(values1).build();
+ Assert.assertEquals(expectedStats, computedStatsObj.getStatsData());
+
+ aggregator.ndvTuner = 0.75;
+ computedStatsObj = aggregator.aggregate(statsList, partitions, true);
+ expectedStats = new ColStatsBuilder<>(Date.class).numNulls(6).numDVs(8)
+ .low(DATE_1).high(DATE_8).fmSketch(values1).build();
+ Assert.assertEquals(expectedStats, computedStatsObj.getStatsData());
+
+ aggregator.ndvTuner = 1;
+ computedStatsObj = aggregator.aggregate(statsList, partitions, true);
+ expectedStats = new ColStatsBuilder<>(Date.class).numNulls(6).numDVs(10)
+ .low(DATE_1).high(DATE_8).fmSketch(values1).build();
+ Assert.assertEquals(expectedStats, computedStatsObj.getStatsData());
+ }
+
+ @Test
+ public void testAggregateMultiStatsWhenOnlySomeAvailable() throws
MetaException {
+ List<String> partitions = Arrays.asList("part1", "part2", "part3",
"part4");
+
+ long[] values1 = { DATE_1.getDaysSinceEpoch(), DATE_2.getDaysSinceEpoch(),
DATE_3.getDaysSinceEpoch() };
+ ColumnStatisticsData data1 = new
ColStatsBuilder<>(Date.class).numNulls(1).numDVs(3)
+ .low(DATE_1).high(DATE_3).hll(values1).build();
+
+ ColumnStatisticsData data3 = new
ColStatsBuilder<>(Date.class).numNulls(3).numDVs(1).low(DATE_7).high(DATE_7)
+ .hll(DATE_7.getDaysSinceEpoch()).build();
+
+ long[] values4 = { DATE_3.getDaysSinceEpoch(), DATE_4.getDaysSinceEpoch(),
DATE_5.getDaysSinceEpoch() };
+ ColumnStatisticsData data4 = new
ColStatsBuilder<>(Date.class).numNulls(2).numDVs(3)
+ .low(DATE_3).high(DATE_5).hll(values4).build();
+
+ List<ColStatsObjWithSourceInfo> statsList =
Arrays.asList(createStatsWithInfo(data1, TABLE, COL, partitions.get(0)),
+ createStatsWithInfo(data3, TABLE, COL, partitions.get(2)),
createStatsWithInfo(data4, TABLE, COL, partitions.get(3)));
+
+ DateColumnStatsAggregator aggregator = new DateColumnStatsAggregator();
+ ColumnStatisticsObj computedStatsObj = aggregator.aggregate(statsList,
partitions, false);
+
+ // hll in case of missing stats is left as null, only numDVs is updated
+ ColumnStatisticsData expectedStats = new
ColStatsBuilder<>(Date.class).numNulls(8).numDVs(4)
+ .low(DATE_1).high(DATE_9).build();
Review Comment:
Also `low` and `high` are computed with interpolation when some stats are
missing, it's an estimated value and it's therefore OK not to be part of the
input in this specific case.
Issue Time Tracking
-------------------
Worklog Id: (was: 808315)
Time Spent: 6h 10m (was: 6h)
> NPEs and rounding issues in ColumnStatsAggregator classes
> ---------------------------------------------------------
>
> Key: HIVE-26277
> URL: https://issues.apache.org/jira/browse/HIVE-26277
> Project: Hive
> Issue Type: Bug
> Components: Standalone Metastore, Statistics, Tests
> Affects Versions: 4.0.0-alpha-2
> Reporter: Alessandro Solimando
> Assignee: Alessandro Solimando
> Priority: Major
> Labels: pull-request-available
> Time Spent: 6h 10m
> Remaining Estimate: 0h
>
> Fix NPEs and rounding errors in _ColumnStatsAggregator_ classes, add
> unit-tests for all the involved classes.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
