Search in sources :

Example 16 with Summary

use of org.apache.accumulo.core.client.summary.Summary in project accumulo by apache.

the class SummaryIT method checkSummaries.

private void checkSummaries(Collection<Summary> summaries, SummarizerConfiguration sc, int total, int missing, int extra, Object... kvs) {
    Summary summary = Iterables.getOnlyElement(summaries);
    Assert.assertEquals("total wrong", total, summary.getFileStatistics().getTotal());
    Assert.assertEquals("missing wrong", missing, summary.getFileStatistics().getMissing());
    Assert.assertEquals("extra wrong", extra, summary.getFileStatistics().getExtra());
    Assert.assertEquals("deleted wrong", 0, summary.getFileStatistics().getDeleted());
    Assert.assertEquals(sc, summary.getSummarizerConfiguration());
    Map<String, Long> expected = new HashMap<>();
    for (int i = 0; i < kvs.length; i += 2) {
        expected.put((String) kvs[i], (Long) kvs[i + 1]);
    }
    Assert.assertEquals(expected, summary.getStatistics());
}
Also used : HashMap(java.util.HashMap) CounterSummary(org.apache.accumulo.core.client.summary.CounterSummary) Summary(org.apache.accumulo.core.client.summary.Summary)

Example 17 with Summary

use of org.apache.accumulo.core.client.summary.Summary 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 18 with Summary

use of org.apache.accumulo.core.client.summary.Summary in project accumulo by apache.

the class SummaryCollection method getSummaries.

public List<Summary> getSummaries() {
    ArrayList<Summary> ret = new ArrayList<>(mergedSummaries.size());
    for (Entry<SummarizerConfiguration, MergedSummary> entry : mergedSummaries.entrySet()) {
        SummarizerConfiguration config = entry.getKey();
        MergedSummary ms = entry.getValue();
        ret.add(new Summary(ms.summary, config, totalFiles, (totalFiles - deletedFiles) - ms.filesContaining, ms.filesExceedingBoundry, ms.filesLarge, deletedFiles));
    }
    return ret;
}
Also used : ArrayList(java.util.ArrayList) Summary(org.apache.accumulo.core.client.summary.Summary) TSummary(org.apache.accumulo.core.data.thrift.TSummary) SummarizerConfiguration(org.apache.accumulo.core.client.summary.SummarizerConfiguration) TSummarizerConfiguration(org.apache.accumulo.core.data.thrift.TSummarizerConfiguration)

Aggregations

Summary (org.apache.accumulo.core.client.summary.Summary)18 SummarizerConfiguration (org.apache.accumulo.core.client.summary.SummarizerConfiguration)12 CounterSummary (org.apache.accumulo.core.client.summary.CounterSummary)11 HashMap (java.util.HashMap)10 Test (org.junit.Test)10 BatchWriter (org.apache.accumulo.core.client.BatchWriter)9 Connector (org.apache.accumulo.core.client.Connector)9 NewTableConfiguration (org.apache.accumulo.core.client.admin.NewTableConfiguration)9 Text (org.apache.hadoop.io.Text)8 BatchWriterConfig (org.apache.accumulo.core.client.BatchWriterConfig)7 ArrayList (java.util.ArrayList)4 Collection (java.util.Collection)4 Map (java.util.Map)4 Entry (java.util.Map.Entry)4 FamilySummarizer (org.apache.accumulo.core.client.summary.summarizers.FamilySummarizer)4 Value (org.apache.accumulo.core.data.Value)4 ImmutableMap (com.google.common.collect.ImmutableMap)3 IOException (java.io.IOException)3 Collections (java.util.Collections)3 List (java.util.List)3