use of org.apache.hadoop.hive.metastore.api.FieldSchema in project hive by apache.
the class TestHBaseStoreIntegration method listPartitionsWithPs.
@Test
public void listPartitionsWithPs() throws Exception {
String dbName = "default";
String tableName = "listPartitionsWithPs";
int startTime = (int) (System.currentTimeMillis() / 1000);
List<FieldSchema> cols = new ArrayList<FieldSchema>();
cols.add(new FieldSchema("col1", "int", "nocomment"));
SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, emptyParameters);
List<FieldSchema> partCols = new ArrayList<FieldSchema>();
partCols.add(new FieldSchema("ds", "string", ""));
partCols.add(new FieldSchema("region", "string", ""));
Table table = new Table(tableName, dbName, "me", startTime, startTime, 0, sd, partCols, emptyParameters, null, null, null);
store.createTable(table);
String[][] partVals = new String[][] { { "today", "north america" }, { "today", "europe" }, { "tomorrow", "north america" }, { "tomorrow", "europe" } };
for (String[] pv : partVals) {
List<String> vals = new ArrayList<String>();
for (String v : pv) vals.add(v);
StorageDescriptor psd = new StorageDescriptor(sd);
psd.setLocation("file:/tmp/ds=" + pv[0] + "/region=" + pv[1]);
Partition part = new Partition(vals, dbName, tableName, startTime, startTime, psd, emptyParameters);
store.addPartition(part);
}
// We only test listPartitionNamesPs since it calls listPartitionsPsWithAuth anyway.
// Test the case where we completely specify the partition
List<String> partitionNames = store.listPartitionNamesPs(dbName, tableName, Arrays.asList(partVals[0]), (short) -1);
Assert.assertEquals(1, partitionNames.size());
Assert.assertEquals("ds=today/region=north america", partitionNames.get(0));
// Leave off the last value of the partition
partitionNames = store.listPartitionNamesPs(dbName, tableName, Arrays.asList(partVals[0][0]), (short) -1);
Assert.assertEquals(2, partitionNames.size());
String[] names = partitionNames.toArray(new String[partitionNames.size()]);
Arrays.sort(names);
Assert.assertArrayEquals(new String[] { "ds=today/region=europe", "ds=today/region=north america" }, names);
// Put a star in the last value of the partition
partitionNames = store.listPartitionNamesPs(dbName, tableName, Arrays.asList("today", "*"), (short) -1);
Assert.assertEquals(2, partitionNames.size());
names = partitionNames.toArray(new String[partitionNames.size()]);
Arrays.sort(names);
Assert.assertArrayEquals(new String[] { "ds=today/region=europe", "ds=today/region=north america" }, names);
// Put a star in the first value of the partition
partitionNames = store.listPartitionNamesPs(dbName, tableName, Arrays.asList("*", "europe"), (short) -1);
Assert.assertEquals(2, partitionNames.size());
names = partitionNames.toArray(new String[partitionNames.size()]);
Arrays.sort(names);
Assert.assertArrayEquals(new String[] { "ds=today/region=europe", "ds=tomorrow/region=europe" }, names);
}
use of org.apache.hadoop.hive.metastore.api.FieldSchema in project hive by apache.
the class TestHBaseAggrStatsCacheIntegration method alterInvalidation.
@Test
public void alterInvalidation() throws Exception {
try {
String dbName = "default";
String tableName = "ai";
List<String> partVals1 = Arrays.asList("today");
List<String> partVals2 = Arrays.asList("yesterday");
List<String> partVals3 = Arrays.asList("tomorrow");
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);
Partition[] partitions = new Partition[3];
int partnum = 0;
for (List<String> partVals : Arrays.asList(partVals1, partVals2, partVals3)) {
StorageDescriptor psd = new StorageDescriptor(sd);
psd.setLocation("file:/tmp/default/invalidation/ds=" + partVals.get(0));
Partition part = new Partition(partVals, dbName, tableName, (int) now, (int) now, psd, Collections.<String, String>emptyMap());
partitions[partnum++] = part;
store.addPartition(part);
ColumnStatistics cs = new ColumnStatistics();
ColumnStatisticsDesc desc = new ColumnStatisticsDesc(false, dbName, tableName);
desc.setLastAnalyzed(now);
desc.setPartName("ds=" + partVals.get(0));
cs.setStatsDesc(desc);
ColumnStatisticsObj obj = new ColumnStatisticsObj();
obj.setColName("col1");
obj.setColType("boolean");
ColumnStatisticsData data = new ColumnStatisticsData();
BooleanColumnStatsData bcsd = new BooleanColumnStatsData();
bcsd.setNumFalses(10);
bcsd.setNumTrues(20);
bcsd.setNumNulls(30);
data.setBooleanStats(bcsd);
obj.setStatsData(data);
cs.addToStatsObj(obj);
store.updatePartitionColumnStatistics(cs, partVals);
}
AggrStats aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=today", "ds=tomorrow"), Arrays.asList("col1"));
aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=today", "ds=yesterday"), Arrays.asList("col1"));
// Check that we had to build it from the stats
Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
Assert.assertEquals(2, store.backdoor().getStatsCache().totalGets.getCnt());
Assert.assertEquals(2, store.backdoor().getStatsCache().misses.getCnt());
// wake the invalidator and check again to make sure it isn't too aggressive about
// removing our stuff.
store.backdoor().getStatsCache().wakeInvalidator();
Partition newPart = new Partition(partitions[2]);
newPart.setLastAccessTime((int) System.currentTimeMillis());
store.alterPartition(dbName, tableName, partVals3, newPart);
store.backdoor().getStatsCache().setRunInvalidatorEvery(100);
store.backdoor().getStatsCache().wakeInvalidator();
aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=tomorrow", "ds=today"), Arrays.asList("col1"));
// Check that we missed, which means this aggregate was dropped from the cache.
Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
Assert.assertEquals(3, store.backdoor().getStatsCache().totalGets.getCnt());
Assert.assertEquals(3, store.backdoor().getStatsCache().misses.getCnt());
// Check that our other aggregate is still in the cache.
aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=yesterday", "ds=today"), Arrays.asList("col1"));
Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
Assert.assertEquals(4, store.backdoor().getStatsCache().totalGets.getCnt());
Assert.assertEquals(3, store.backdoor().getStatsCache().misses.getCnt());
} finally {
store.backdoor().getStatsCache().setRunInvalidatorEvery(5000);
store.backdoor().getStatsCache().setMaxTimeInCache(500000);
store.backdoor().getStatsCache().wakeInvalidator();
}
}
use of org.apache.hadoop.hive.metastore.api.FieldSchema in project hive by apache.
the class TestHBaseStoreIntegration method alterPartitions.
@Test
public void alterPartitions() throws Exception {
String dbName = "default";
String tableName = "alterParts";
int startTime = (int) (System.currentTimeMillis() / 1000);
List<FieldSchema> cols = new ArrayList<FieldSchema>();
cols.add(new FieldSchema("col1", "int", "nocomment"));
SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, emptyParameters);
List<FieldSchema> partCols = new ArrayList<FieldSchema>();
partCols.add(new FieldSchema("pc", "string", ""));
Table table = new Table(tableName, dbName, "me", startTime, startTime, 0, sd, partCols, emptyParameters, null, null, null);
store.createTable(table);
List<String> partVals = Arrays.asList("alan", "bob", "carl", "doug", "ethan");
List<Partition> partitions = new ArrayList<Partition>();
List<List<String>> allVals = new ArrayList<List<String>>();
for (String val : partVals) {
List<String> vals = new ArrayList<String>();
allVals.add(vals);
vals.add(val);
StorageDescriptor psd = new StorageDescriptor(sd);
psd.setLocation("file:/tmp/pc=" + val);
Partition part = new Partition(vals, dbName, tableName, startTime, startTime, psd, emptyParameters);
partitions.add(part);
}
store.addPartitions(dbName, tableName, partitions);
for (Partition p : partitions) p.setLastAccessTime(startTime + 10);
store.alterPartitions(dbName, tableName, allVals, partitions);
partitions = store.getPartitions(dbName, tableName, -1);
for (Partition part : partitions) {
Assert.assertEquals(startTime + 10, part.getLastAccessTime());
}
}
use of org.apache.hadoop.hive.metastore.api.FieldSchema in project hive by apache.
the class TestHBaseStoreIntegration method getPartitions.
@Test
public void getPartitions() throws Exception {
String dbName = "default";
String tableName = "manyParts";
int startTime = (int) (System.currentTimeMillis() / 1000);
List<FieldSchema> cols = new ArrayList<FieldSchema>();
cols.add(new FieldSchema("col1", "int", "nocomment"));
SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, emptyParameters);
List<FieldSchema> partCols = new ArrayList<FieldSchema>();
partCols.add(new FieldSchema("pc", "string", ""));
Table table = new Table(tableName, dbName, "me", startTime, startTime, 0, sd, partCols, emptyParameters, null, null, null);
store.createTable(table);
List<String> partVals = Arrays.asList("alan", "bob", "carl", "doug", "ethan");
for (String val : partVals) {
List<String> vals = new ArrayList<String>();
vals.add(val);
StorageDescriptor psd = new StorageDescriptor(sd);
psd.setLocation("file:/tmp/pc=" + val);
Partition part = new Partition(vals, dbName, tableName, startTime, startTime, psd, emptyParameters);
store.addPartition(part);
Partition p = store.getPartition(dbName, tableName, vals);
Assert.assertEquals("file:/tmp/pc=" + val, p.getSd().getLocation());
}
List<Partition> parts = store.getPartitions(dbName, tableName, -1);
Assert.assertEquals(5, parts.size());
String[] pv = new String[5];
for (int i = 0; i < 5; i++) pv[i] = parts.get(i).getValues().get(0);
Arrays.sort(pv);
Assert.assertArrayEquals(pv, partVals.toArray(new String[5]));
}
use of org.apache.hadoop.hive.metastore.api.FieldSchema in project hive by apache.
the class TestHBaseStoreIntegration method addPartitions.
@Test
public void addPartitions() throws Exception {
String dbName = "default";
String tableName = "addParts";
int startTime = (int) (System.currentTimeMillis() / 1000);
List<FieldSchema> cols = new ArrayList<FieldSchema>();
cols.add(new FieldSchema("col1", "int", "nocomment"));
SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, emptyParameters);
List<FieldSchema> partCols = new ArrayList<FieldSchema>();
partCols.add(new FieldSchema("pc", "string", ""));
Table table = new Table(tableName, dbName, "me", startTime, startTime, 0, sd, partCols, emptyParameters, null, null, null);
store.createTable(table);
List<String> partVals = Arrays.asList("alan", "bob", "carl", "doug", "ethan");
List<Partition> partitions = new ArrayList<Partition>();
for (String val : partVals) {
List<String> vals = new ArrayList<String>();
vals.add(val);
StorageDescriptor psd = new StorageDescriptor(sd);
psd.setLocation("file:/tmp/pc=" + val);
Partition part = new Partition(vals, dbName, tableName, startTime, startTime, psd, emptyParameters);
partitions.add(part);
}
store.addPartitions(dbName, tableName, partitions);
List<String> partNames = store.listPartitionNames(dbName, tableName, (short) -1);
Assert.assertEquals(5, partNames.size());
String[] names = partNames.toArray(new String[partNames.size()]);
Arrays.sort(names);
String[] canonicalNames = partVals.toArray(new String[partVals.size()]);
for (int i = 0; i < canonicalNames.length; i++) canonicalNames[i] = "pc=" + canonicalNames[i];
Assert.assertArrayEquals(canonicalNames, names);
}
Aggregations