Search in sources :

Example 6 with CompactionStrategyConfig

use of org.apache.accumulo.core.client.admin.CompactionStrategyConfig in project accumulo by apache.

the class UserCompactionStrategyIT method testDropA.

@Test
public void testDropA() throws Exception {
    Connector c = getConnector();
    String tableName = getUniqueNames(1)[0];
    c.tableOperations().create(tableName);
    writeFlush(c, tableName, "a");
    writeFlush(c, tableName, "b");
    // create a file that starts with A containing rows 'a' and 'b'
    c.tableOperations().compact(tableName, new CompactionConfig().setWait(true));
    writeFlush(c, tableName, "c");
    writeFlush(c, tableName, "d");
    // drop files that start with A
    CompactionStrategyConfig csConfig = new CompactionStrategyConfig(TestCompactionStrategy.class.getName());
    csConfig.setOptions(ImmutableMap.of("dropPrefix", "A", "inputPrefix", "F"));
    c.tableOperations().compact(tableName, new CompactionConfig().setWait(true).setCompactionStrategy(csConfig));
    Assert.assertEquals(ImmutableSet.of("c", "d"), getRows(c, tableName));
    // this compaction should not drop files starting with A
    c.tableOperations().compact(tableName, new CompactionConfig().setWait(true));
    c.tableOperations().compact(tableName, new CompactionConfig().setWait(true));
    Assert.assertEquals(ImmutableSet.of("c", "d"), getRows(c, tableName));
}
Also used : CompactionStrategyConfig(org.apache.accumulo.core.client.admin.CompactionStrategyConfig) Connector(org.apache.accumulo.core.client.Connector) CompactionConfig(org.apache.accumulo.core.client.admin.CompactionConfig) Test(org.junit.Test)

Example 7 with CompactionStrategyConfig

use of org.apache.accumulo.core.client.admin.CompactionStrategyConfig in project accumulo by apache.

the class UserCompactionStrategyIT method testDropNone.

private void testDropNone(Map<String, String> options) throws Exception {
    Connector c = getConnector();
    String tableName = getUniqueNames(1)[0];
    c.tableOperations().create(tableName);
    writeFlush(c, tableName, "a");
    writeFlush(c, tableName, "b");
    CompactionStrategyConfig csConfig = new CompactionStrategyConfig(TestCompactionStrategy.class.getName());
    csConfig.setOptions(options);
    c.tableOperations().compact(tableName, new CompactionConfig().setWait(true).setCompactionStrategy(csConfig));
    Assert.assertEquals(ImmutableSet.of("a", "b"), getRows(c, tableName));
}
Also used : CompactionStrategyConfig(org.apache.accumulo.core.client.admin.CompactionStrategyConfig) Connector(org.apache.accumulo.core.client.Connector) CompactionConfig(org.apache.accumulo.core.client.admin.CompactionConfig)

Example 8 with CompactionStrategyConfig

use of org.apache.accumulo.core.client.admin.CompactionStrategyConfig in project accumulo by apache.

the class SummaryIT method compactionTest.

@Test
public void compactionTest() throws Exception {
    final String table = getUniqueNames(1)[0];
    Connector c = getConnector();
    NewTableConfiguration ntc = new NewTableConfiguration();
    SummarizerConfiguration sc1 = SummarizerConfiguration.builder(FooCounter.class.getName()).build();
    ntc.enableSummarization(sc1);
    c.tableOperations().create(table, ntc);
    try (BatchWriter bw = c.createBatchWriter(table, new BatchWriterConfig())) {
        write(bw, "bar1", "f1", "q1", "v1");
        write(bw, "bar2", "f1", "q1", "v2");
        write(bw, "foo1", "f1", "q1", "v3");
    }
    // Create a compaction config that will filter out foos if there are too many. Uses summary data to know if there are too many foos.
    CompactionStrategyConfig csc = new CompactionStrategyConfig(FooCS.class.getName());
    List<IteratorSetting> iterators = Collections.singletonList(new IteratorSetting(100, FooFilter.class));
    CompactionConfig compactConfig = new CompactionConfig().setFlush(true).setCompactionStrategy(csc).setIterators(iterators).setWait(true);
    // this compaction should make no changes because there are less foos than bars
    c.tableOperations().compact(table, compactConfig);
    try (Scanner scanner = c.createScanner(table, Authorizations.EMPTY)) {
        Stream<Entry<Key, Value>> stream = StreamSupport.stream(scanner.spliterator(), false);
        Map<String, Long> counts = // convert to row
        stream.map(e -> e.getKey().getRowData().toString()).map(// strip numbers off row
        r -> r.replaceAll("[0-9]+", "")).collect(// count different row types
        groupingBy(identity(), counting()));
        Assert.assertEquals(1l, (long) counts.getOrDefault("foo", 0l));
        Assert.assertEquals(2l, (long) counts.getOrDefault("bar", 0l));
        Assert.assertEquals(2, counts.size());
    }
    try (BatchWriter bw = c.createBatchWriter(table, new BatchWriterConfig())) {
        write(bw, "foo2", "f1", "q1", "v4");
        write(bw, "foo3", "f1", "q1", "v5");
        write(bw, "foo4", "f1", "q1", "v6");
    }
    // this compaction should remove all foos because there are more foos than bars
    c.tableOperations().compact(table, compactConfig);
    try (Scanner scanner = c.createScanner(table, Authorizations.EMPTY)) {
        Stream<Entry<Key, Value>> stream = StreamSupport.stream(scanner.spliterator(), false);
        Map<String, Long> counts = // convert to row
        stream.map(e -> e.getKey().getRowData().toString()).map(// strip numbers off row
        r -> r.replaceAll("[0-9]+", "")).collect(// count different row types
        groupingBy(identity(), counting()));
        Assert.assertEquals(0l, (long) counts.getOrDefault("foo", 0l));
        Assert.assertEquals(2l, (long) counts.getOrDefault("bar", 0l));
        Assert.assertEquals(1, counts.size());
    }
}
Also used : Arrays(java.util.Arrays) TableOfflineException(org.apache.accumulo.core.client.TableOfflineException) SortedSet(java.util.SortedSet) Collectors.counting(java.util.stream.Collectors.counting) DELETES_IGNORED_STAT(org.apache.accumulo.core.client.summary.CountingSummarizer.DELETES_IGNORED_STAT) TOO_MANY_STAT(org.apache.accumulo.core.client.summary.CountingSummarizer.TOO_MANY_STAT) AccumuloServerException(org.apache.accumulo.core.client.impl.AccumuloServerException) Text(org.apache.hadoop.io.Text) Random(java.util.Random) Mutation(org.apache.accumulo.core.data.Mutation) TableNotFoundException(org.apache.accumulo.core.client.TableNotFoundException) MIN_TIMESTAMP_STAT(org.apache.accumulo.test.functional.BasicSummarizer.MIN_TIMESTAMP_STAT) BatchWriterConfig(org.apache.accumulo.core.client.BatchWriterConfig) Summarizer(org.apache.accumulo.core.client.summary.Summarizer) TOTAL_STAT(org.apache.accumulo.test.functional.BasicSummarizer.TOTAL_STAT) Map(java.util.Map) Value(org.apache.accumulo.core.data.Value) PatternSyntaxException(java.util.regex.PatternSyntaxException) ImmutableSet(com.google.common.collect.ImmutableSet) ImmutableMap(com.google.common.collect.ImmutableMap) TOO_LONG_STAT(org.apache.accumulo.core.client.summary.CountingSummarizer.TOO_LONG_STAT) Collection(java.util.Collection) MutationsRejectedException(org.apache.accumulo.core.client.MutationsRejectedException) Set(java.util.Set) SecurityErrorCode(org.apache.accumulo.core.client.security.SecurityErrorCode) CompactionStrategyConfig(org.apache.accumulo.core.client.admin.CompactionStrategyConfig) List(java.util.List) PasswordToken(org.apache.accumulo.core.client.security.tokens.PasswordToken) Filter(org.apache.accumulo.core.iterators.Filter) Stream(java.util.stream.Stream) FamilySummarizer(org.apache.accumulo.core.client.summary.summarizers.FamilySummarizer) DELETES_STAT(org.apache.accumulo.test.functional.BasicSummarizer.DELETES_STAT) Function.identity(java.util.function.Function.identity) Entry(java.util.Map.Entry) Scanner(org.apache.accumulo.core.client.Scanner) ZooKeeperInstance(org.apache.accumulo.core.client.ZooKeeperInstance) SEEN_STAT(org.apache.accumulo.core.client.summary.CountingSummarizer.SEEN_STAT) CompactionPlan(org.apache.accumulo.tserver.compaction.CompactionPlan) Iterables(com.google.common.collect.Iterables) VisibilitySummarizer(org.apache.accumulo.core.client.summary.summarizers.VisibilitySummarizer) SummarizerConfiguration(org.apache.accumulo.core.client.summary.SummarizerConfiguration) UtilWaitThread(org.apache.accumulo.fate.util.UtilWaitThread) Collectors.groupingBy(java.util.stream.Collectors.groupingBy) FileStatistics(org.apache.accumulo.core.client.summary.Summary.FileStatistics) HashMap(java.util.HashMap) CompactionConfig(org.apache.accumulo.core.client.admin.CompactionConfig) TreeSet(java.util.TreeSet) Connector(org.apache.accumulo.core.client.Connector) Builder(com.google.common.collect.ImmutableMap.Builder) NewTableConfiguration(org.apache.accumulo.core.client.admin.NewTableConfiguration) AccumuloClusterHarness(org.apache.accumulo.harness.AccumuloClusterHarness) TablePermission(org.apache.accumulo.core.security.TablePermission) Lists(com.google.common.collect.Lists) AccumuloSecurityException(org.apache.accumulo.core.client.AccumuloSecurityException) Key(org.apache.accumulo.core.data.Key) EMITTED_STAT(org.apache.accumulo.core.client.summary.CountingSummarizer.EMITTED_STAT) MAX_TIMESTAMP_STAT(org.apache.accumulo.test.functional.BasicSummarizer.MAX_TIMESTAMP_STAT) StreamSupport(java.util.stream.StreamSupport) CounterSummary(org.apache.accumulo.core.client.summary.CounterSummary) LongSummaryStatistics(java.util.LongSummaryStatistics) Summary(org.apache.accumulo.core.client.summary.Summary) MajorCompactionRequest(org.apache.accumulo.tserver.compaction.MajorCompactionRequest) IOException(java.io.IOException) Test(org.junit.Test) Authorizations(org.apache.accumulo.core.security.Authorizations) AccumuloException(org.apache.accumulo.core.client.AccumuloException) Range(org.apache.accumulo.core.data.Range) IteratorSetting(org.apache.accumulo.core.client.IteratorSetting) BatchWriter(org.apache.accumulo.core.client.BatchWriter) CompactionStrategy(org.apache.accumulo.tserver.compaction.CompactionStrategy) Assert(org.junit.Assert) Collections(java.util.Collections) Connector(org.apache.accumulo.core.client.Connector) Scanner(org.apache.accumulo.core.client.Scanner) CompactionStrategyConfig(org.apache.accumulo.core.client.admin.CompactionStrategyConfig) Entry(java.util.Map.Entry) IteratorSetting(org.apache.accumulo.core.client.IteratorSetting) NewTableConfiguration(org.apache.accumulo.core.client.admin.NewTableConfiguration) CompactionConfig(org.apache.accumulo.core.client.admin.CompactionConfig) BatchWriterConfig(org.apache.accumulo.core.client.BatchWriterConfig) BatchWriter(org.apache.accumulo.core.client.BatchWriter) SummarizerConfiguration(org.apache.accumulo.core.client.summary.SummarizerConfiguration) Test(org.junit.Test)

Example 9 with CompactionStrategyConfig

use of org.apache.accumulo.core.client.admin.CompactionStrategyConfig in project accumulo by apache.

the class CompactionStrategyConfigUtil method decode.

public static CompactionStrategyConfig decode(DataInput din) throws IOException {
    if (din.readInt() != MAGIC) {
        throw new IllegalArgumentException("Unexpected MAGIC ");
    }
    if (din.readByte() != 1) {
        throw new IllegalArgumentException("Unexpected version");
    }
    String classname = din.readUTF();
    int numEntries = din.readInt();
    HashMap<String, String> opts = new HashMap<>();
    for (int i = 0; i < numEntries; i++) {
        String k = din.readUTF();
        String v = din.readUTF();
        opts.put(k, v);
    }
    return new CompactionStrategyConfig(classname).setOptions(opts);
}
Also used : CompactionStrategyConfig(org.apache.accumulo.core.client.admin.CompactionStrategyConfig) HashMap(java.util.HashMap)

Aggregations

CompactionStrategyConfig (org.apache.accumulo.core.client.admin.CompactionStrategyConfig)9 CompactionConfig (org.apache.accumulo.core.client.admin.CompactionConfig)7 Connector (org.apache.accumulo.core.client.Connector)6 Test (org.junit.Test)5 HashMap (java.util.HashMap)3 IteratorSetting (org.apache.accumulo.core.client.IteratorSetting)3 IOException (java.io.IOException)2 CompactionStrategy (org.apache.accumulo.tserver.compaction.CompactionStrategy)2 MajorCompactionRequest (org.apache.accumulo.tserver.compaction.MajorCompactionRequest)2 ImmutableMap (com.google.common.collect.ImmutableMap)1 Builder (com.google.common.collect.ImmutableMap.Builder)1 ImmutableSet (com.google.common.collect.ImmutableSet)1 Iterables (com.google.common.collect.Iterables)1 Lists (com.google.common.collect.Lists)1 File (java.io.File)1 Arrays (java.util.Arrays)1 Collection (java.util.Collection)1 Collections (java.util.Collections)1 List (java.util.List)1 LongSummaryStatistics (java.util.LongSummaryStatistics)1