use of org.apache.hadoop.hive.metastore.api.AggrStats in project hive by apache.
the class TestHBaseAggregateStatsCache method noneWithStats.
@Test
public void noneWithStats() throws Exception {
String dbName = "default";
String tableName = "nws";
List<String> partVals1 = Arrays.asList("today");
List<String> partVals2 = Arrays.asList("yesterday");
long now = System.currentTimeMillis();
List<FieldSchema> cols = new ArrayList<>();
cols.add(new FieldSchema("col1", "boolean", "nocomment"));
SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, Collections.<String, String>emptyMap());
List<FieldSchema> partCols = new ArrayList<>();
partCols.add(new FieldSchema("ds", "string", ""));
Table table = new Table(tableName, dbName, "me", (int) now, (int) now, 0, sd, partCols, Collections.<String, String>emptyMap(), null, null, null);
store.createTable(table);
for (List<String> partVals : Arrays.asList(partVals1, partVals2)) {
StorageDescriptor psd = new StorageDescriptor(sd);
psd.setLocation("file:/tmp/default/nws/ds=" + partVals.get(0));
Partition part = new Partition(partVals, dbName, tableName, (int) now, (int) now, psd, Collections.<String, String>emptyMap());
store.addPartition(part);
}
Checker statChecker = new Checker() {
@Override
public void checkStats(AggrStats aggrStats) throws Exception {
Assert.assertEquals(0, aggrStats.getPartsFound());
}
};
AggrStats aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=today", "ds=yesterday"), Arrays.asList("col1"));
statChecker.checkStats(aggrStats);
}
use of org.apache.hadoop.hive.metastore.api.AggrStats in project hive by apache.
the class TestHBaseAggregateStatsCacheWithBitVector method allPartitions.
@Test
public void allPartitions() throws Exception {
String dbName = "default";
String tableName = "snp";
List<String> partVals1 = Arrays.asList("today");
List<String> partVals2 = Arrays.asList("yesterday");
long now = System.currentTimeMillis();
List<FieldSchema> cols = new ArrayList<>();
cols.add(new FieldSchema("col1", "boolean", "nocomment"));
SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, Collections.<String, String>emptyMap());
List<FieldSchema> partCols = new ArrayList<>();
partCols.add(new FieldSchema("ds", "string", ""));
Table table = new Table(tableName, dbName, "me", (int) now, (int) now, 0, sd, partCols, Collections.<String, String>emptyMap(), null, null, null);
store.createTable(table);
StorageDescriptor psd = new StorageDescriptor(sd);
psd.setLocation("file:/tmp/default/hit/ds=" + partVals1.get(0));
Partition part = new Partition(partVals1, dbName, tableName, (int) now, (int) now, psd, Collections.<String, String>emptyMap());
store.addPartition(part);
psd = new StorageDescriptor(sd);
psd.setLocation("file:/tmp/default/hit/ds=" + partVals2.get(0));
part = new Partition(partVals2, dbName, tableName, (int) now, (int) now, psd, Collections.<String, String>emptyMap());
store.addPartition(part);
ColumnStatistics cs = new ColumnStatistics();
ColumnStatisticsDesc desc = new ColumnStatisticsDesc(false, dbName, tableName);
desc.setLastAnalyzed(now);
desc.setPartName("ds=" + partVals1.get(0));
cs.setStatsDesc(desc);
ColumnStatisticsObj obj = new ColumnStatisticsObj();
obj.setColName("col1");
obj.setColType("double");
ColumnStatisticsData data = new ColumnStatisticsData();
DoubleColumnStatsData dcsd = new DoubleColumnStatsData();
dcsd.setHighValue(1000.2342343);
dcsd.setLowValue(-20.1234213423);
dcsd.setNumNulls(30);
dcsd.setNumDVs(12342);
dcsd.setBitVectors("{0, 4, 5, 7}{0, 1}{0, 1, 2}{0, 1, 4}{0}{0, 2}{0, 3}{0, 2, 3, 4}{0, 1, 4}{0, 1}{0}{0, 1, 3, 8}{0, 2}{0, 2}{0, 9}{0, 1, 4}");
data.setDoubleStats(dcsd);
obj.setStatsData(data);
cs.addToStatsObj(obj);
store.updatePartitionColumnStatistics(cs, partVals1);
cs = new ColumnStatistics();
desc = new ColumnStatisticsDesc(false, dbName, tableName);
desc.setLastAnalyzed(now);
desc.setPartName("ds=" + partVals2.get(0));
cs.setStatsDesc(desc);
obj = new ColumnStatisticsObj();
obj.setColName("col1");
obj.setColType("double");
data = new ColumnStatisticsData();
dcsd = new DoubleColumnStatsData();
dcsd.setHighValue(1000.2342343);
dcsd.setLowValue(-20.1234213423);
dcsd.setNumNulls(30);
dcsd.setNumDVs(12342);
dcsd.setBitVectors("{0, 1}{0, 1}{1, 2, 4}{0, 1, 2}{0, 1, 2}{0, 2}{0, 1, 3, 4}{0, 1}{0, 1}{3, 4, 6}{2}{0, 1}{0, 3}{0}{0, 1}{0, 1, 4}");
data.setDoubleStats(dcsd);
obj.setStatsData(data);
cs.addToStatsObj(obj);
store.updatePartitionColumnStatistics(cs, partVals2);
Checker statChecker = new Checker() {
@Override
public void checkStats(AggrStats aggrStats) throws Exception {
Assert.assertEquals(2, aggrStats.getPartsFound());
Assert.assertEquals(1, aggrStats.getColStatsSize());
ColumnStatisticsObj cso = aggrStats.getColStats().get(0);
Assert.assertEquals("col1", cso.getColName());
Assert.assertEquals("double", cso.getColType());
DoubleColumnStatsData dcsd = cso.getStatsData().getDoubleStats();
Assert.assertEquals(1000.23, dcsd.getHighValue(), 0.01);
Assert.assertEquals(-20.12, dcsd.getLowValue(), 0.01);
Assert.assertEquals(60, dcsd.getNumNulls());
Assert.assertEquals(5, dcsd.getNumDVs());
// we do not store the bitvector for the aggrStats.
// we can store that if it is necessary in the future.
}
};
AggrStats aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=today", "ds=yesterday"), Arrays.asList("col1"));
statChecker.checkStats(aggrStats);
// Check that we had to build it from the stats
Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
Assert.assertEquals(1, store.backdoor().getStatsCache().totalGets.getCnt());
Assert.assertEquals(1, store.backdoor().getStatsCache().misses.getCnt());
// Call again, this time it should come from memory. Also, reverse the name
// order this time
// to assure that we still hit.
aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=yesterday", "ds=today"), Arrays.asList("col1"));
statChecker.checkStats(aggrStats);
Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
Assert.assertEquals(2, store.backdoor().getStatsCache().totalGets.getCnt());
Assert.assertEquals(1, store.backdoor().getStatsCache().misses.getCnt());
}
use of org.apache.hadoop.hive.metastore.api.AggrStats in project hive by apache.
the class TestHBaseAggregateStatsExtrapolation method TwoEndsOfPartitionsHaveBitVectorStatus.
@Test
public void TwoEndsOfPartitionsHaveBitVectorStatus() throws Exception {
String dbName = "default";
String tableName = "snp";
long now = System.currentTimeMillis();
List<FieldSchema> cols = new ArrayList<>();
cols.add(new FieldSchema("col3", "long", "nocomment"));
SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, Collections.<String, String>emptyMap());
List<FieldSchema> partCols = new ArrayList<>();
partCols.add(new FieldSchema("ds", "string", ""));
Table table = new Table(tableName, dbName, "me", (int) now, (int) now, 0, sd, partCols, Collections.<String, String>emptyMap(), null, null, null);
store.createTable(table);
List<List<String>> partVals = new ArrayList<>();
for (int i = 0; i < 10; i++) {
List<String> partVal = Arrays.asList("" + i);
partVals.add(partVal);
StorageDescriptor psd = new StorageDescriptor(sd);
psd.setLocation("file:/tmp/default/hit/ds=" + partVal);
Partition part = new Partition(partVal, dbName, tableName, (int) now, (int) now, psd, Collections.<String, String>emptyMap());
store.addPartition(part);
if (i < 2 || i > 7) {
ColumnStatistics cs = new ColumnStatistics();
ColumnStatisticsDesc desc = new ColumnStatisticsDesc(false, dbName, tableName);
desc.setLastAnalyzed(now);
desc.setPartName("ds=" + partVal);
cs.setStatsDesc(desc);
ColumnStatisticsObj obj = new ColumnStatisticsObj();
obj.setColName("col3");
obj.setColType("long");
ColumnStatisticsData data = new ColumnStatisticsData();
LongColumnStatsData dcsd = new LongColumnStatsData();
dcsd.setHighValue(1000 + i);
dcsd.setLowValue(-1000 - i);
dcsd.setNumNulls(i);
dcsd.setNumDVs(10 * i);
dcsd.setBitVectors(bitVectors);
data.setLongStats(dcsd);
obj.setStatsData(data);
cs.addToStatsObj(obj);
store.updatePartitionColumnStatistics(cs, partVal);
}
}
Checker statChecker = new Checker() {
@Override
public void checkStats(AggrStats aggrStats) throws Exception {
Assert.assertEquals(4, aggrStats.getPartsFound());
Assert.assertEquals(1, aggrStats.getColStatsSize());
ColumnStatisticsObj cso = aggrStats.getColStats().get(0);
Assert.assertEquals("col3", cso.getColName());
Assert.assertEquals("long", cso.getColType());
LongColumnStatsData lcsd = cso.getStatsData().getLongStats();
Assert.assertEquals(1010, lcsd.getHighValue(), 0.01);
Assert.assertEquals(-1010, lcsd.getLowValue(), 0.01);
Assert.assertEquals(45, lcsd.getNumNulls());
Assert.assertEquals(3, lcsd.getNumDVs());
}
};
List<String> partNames = new ArrayList<>();
for (int i = 0; i < 10; i++) {
partNames.add("ds=" + i);
}
AggrStats aggrStats = store.get_aggr_stats_for(dbName, tableName, partNames, Arrays.asList("col3"));
statChecker.checkStats(aggrStats);
}
use of org.apache.hadoop.hive.metastore.api.AggrStats in project hive by apache.
the class TestHBaseAggregateStatsExtrapolation method noPartitionsHaveBitVectorStatus.
@Test
public void noPartitionsHaveBitVectorStatus() throws Exception {
String dbName = "default";
String tableName = "snp";
long now = System.currentTimeMillis();
List<FieldSchema> cols = new ArrayList<>();
cols.add(new FieldSchema("col2", "long", "nocomment"));
SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, Collections.<String, String>emptyMap());
List<FieldSchema> partCols = new ArrayList<>();
partCols.add(new FieldSchema("ds", "string", ""));
Table table = new Table(tableName, dbName, "me", (int) now, (int) now, 0, sd, partCols, Collections.<String, String>emptyMap(), null, null, null);
store.createTable(table);
List<List<String>> partVals = new ArrayList<>();
for (int i = 0; i < 10; i++) {
List<String> partVal = Arrays.asList("" + i);
partVals.add(partVal);
StorageDescriptor psd = new StorageDescriptor(sd);
psd.setLocation("file:/tmp/default/hit/ds=" + partVal);
Partition part = new Partition(partVal, dbName, tableName, (int) now, (int) now, psd, Collections.<String, String>emptyMap());
store.addPartition(part);
ColumnStatistics cs = new ColumnStatistics();
ColumnStatisticsDesc desc = new ColumnStatisticsDesc(false, dbName, tableName);
desc.setLastAnalyzed(now);
desc.setPartName("ds=" + partVal);
cs.setStatsDesc(desc);
ColumnStatisticsObj obj = new ColumnStatisticsObj();
obj.setColName("col2");
obj.setColType("long");
ColumnStatisticsData data = new ColumnStatisticsData();
LongColumnStatsData dcsd = new LongColumnStatsData();
dcsd.setHighValue(1000 + i);
dcsd.setLowValue(-1000 - i);
dcsd.setNumNulls(i);
dcsd.setNumDVs(10 * i);
data.setLongStats(dcsd);
obj.setStatsData(data);
cs.addToStatsObj(obj);
store.updatePartitionColumnStatistics(cs, partVal);
}
Checker statChecker = new Checker() {
@Override
public void checkStats(AggrStats aggrStats) throws Exception {
Assert.assertEquals(10, aggrStats.getPartsFound());
Assert.assertEquals(1, aggrStats.getColStatsSize());
ColumnStatisticsObj cso = aggrStats.getColStats().get(0);
Assert.assertEquals("col2", cso.getColName());
Assert.assertEquals("long", cso.getColType());
LongColumnStatsData lcsd = cso.getStatsData().getLongStats();
Assert.assertEquals(1009, lcsd.getHighValue(), 0.01);
Assert.assertEquals(-1009, lcsd.getLowValue(), 0.01);
Assert.assertEquals(45, lcsd.getNumNulls());
Assert.assertEquals(90, lcsd.getNumDVs());
}
};
List<String> partNames = new ArrayList<>();
for (int i = 0; i < 10; i++) {
partNames.add("ds=" + i);
}
AggrStats aggrStats = store.get_aggr_stats_for(dbName, tableName, partNames, Arrays.asList("col2"));
statChecker.checkStats(aggrStats);
}
use of org.apache.hadoop.hive.metastore.api.AggrStats in project hive by apache.
the class HBaseStore method get_aggr_stats_for.
/**
* Return aggregated statistics for each column in the colNames list aggregated over partitions in
* the partNames list
*
*/
@Override
public AggrStats get_aggr_stats_for(String dbName, String tblName, List<String> partNames, List<String> colNames) throws MetaException, NoSuchObjectException {
List<List<String>> partVals = new ArrayList<List<String>>(partNames.size());
for (String partName : partNames) {
partVals.add(partNameToVals(partName));
}
boolean commit = false;
boolean hasAnyStats = false;
openTransaction();
try {
AggrStats aggrStats = new AggrStats();
aggrStats.setPartsFound(0);
for (String colName : colNames) {
try {
AggrStats oneCol = getHBase().getStatsCache().get(dbName, tblName, partNames, colName);
if (oneCol.getColStatsSize() > 0) {
assert oneCol.getColStatsSize() == 1;
aggrStats.setPartsFound(oneCol.getPartsFound());
aggrStats.addToColStats(oneCol.getColStats().get(0));
hasAnyStats = true;
}
} catch (CacheLoader.InvalidCacheLoadException e) {
LOG.debug("Found no stats for column " + colName);
// This means we have no stats at all for this column for these partitions, so just
// move on.
}
}
commit = true;
if (!hasAnyStats) {
// Set the required field.
aggrStats.setColStats(new ArrayList<ColumnStatisticsObj>());
}
return aggrStats;
} catch (IOException e) {
LOG.error("Unable to fetch aggregate column statistics", e);
throw new MetaException("Failed fetching aggregate column statistics, " + e.getMessage());
} finally {
commitOrRoleBack(commit);
}
}
Aggregations