Search in sources :

Example 6 with CounterSummary

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

the class SummaryIT method testManyFiles.

@Test
public void testManyFiles() throws Exception {
    final String table = getUniqueNames(1)[0];
    Connector c = getConnector();
    NewTableConfiguration ntc = new NewTableConfiguration();
    ntc.enableSummarization(SummarizerConfiguration.builder(FamilySummarizer.class).build());
    c.tableOperations().create(table, ntc);
    Random rand = new Random(42);
    int q = 0;
    SortedSet<Text> partitionKeys = new TreeSet<>();
    for (int split = 100_000; split < 1_000_000; split += 100_000) {
        partitionKeys.add(new Text(String.format("%06d", split)));
    }
    c.tableOperations().addSplits(table, partitionKeys);
    Map<String, Long> famCounts = new HashMap<>();
    for (int t = 0; t < 20; t++) {
        // this loop should cause a varying number of files and compactions
        try (BatchWriter bw = c.createBatchWriter(table, new BatchWriterConfig())) {
            for (int i = 0; i < 10000; i++) {
                String row = String.format("%06d", rand.nextInt(1_000_000));
                String fam = String.format("%03d", rand.nextInt(100));
                String qual = String.format("%06d", q++);
                write(bw, row, fam, qual, "val");
                famCounts.merge(fam, 1L, Long::sum);
            }
        }
        List<Summary> summaries = c.tableOperations().summaries(table).flush(true).retrieve();
        Assert.assertEquals(1, summaries.size());
        CounterSummary cs = new CounterSummary(summaries.get(0));
        Assert.assertEquals(famCounts, cs.getCounters());
        FileStatistics fileStats = summaries.get(0).getFileStatistics();
        Assert.assertEquals(0, fileStats.getInaccurate());
        Assert.assertTrue("Saw " + fileStats.getTotal() + " files expected >=10", fileStats.getTotal() >= 10);
    }
}
Also used : Connector(org.apache.accumulo.core.client.Connector) FileStatistics(org.apache.accumulo.core.client.summary.Summary.FileStatistics) HashMap(java.util.HashMap) Text(org.apache.hadoop.io.Text) Random(java.util.Random) NewTableConfiguration(org.apache.accumulo.core.client.admin.NewTableConfiguration) TreeSet(java.util.TreeSet) CounterSummary(org.apache.accumulo.core.client.summary.CounterSummary) BatchWriterConfig(org.apache.accumulo.core.client.BatchWriterConfig) CounterSummary(org.apache.accumulo.core.client.summary.CounterSummary) Summary(org.apache.accumulo.core.client.summary.Summary) BatchWriter(org.apache.accumulo.core.client.BatchWriter) Test(org.junit.Test)

Aggregations

CounterSummary (org.apache.accumulo.core.client.summary.CounterSummary)6 HashMap (java.util.HashMap)5 Test (org.junit.Test)5 SummarizerConfiguration (org.apache.accumulo.core.client.summary.SummarizerConfiguration)4 FamilySummarizer (org.apache.accumulo.core.client.summary.summarizers.FamilySummarizer)4 VisibilitySummarizer (org.apache.accumulo.core.client.summary.summarizers.VisibilitySummarizer)4 Key (org.apache.accumulo.core.data.Key)4 Value (org.apache.accumulo.core.data.Value)4 Collector (org.apache.accumulo.core.client.summary.Summarizer.Collector)3 Summary (org.apache.accumulo.core.client.summary.Summary)3 Random (java.util.Random)2 NewTableConfiguration (org.apache.accumulo.core.client.admin.NewTableConfiguration)2 CountingSummarizer (org.apache.accumulo.core.client.summary.CountingSummarizer)2 Summarizer (org.apache.accumulo.core.client.summary.Summarizer)2 Text (org.apache.hadoop.io.Text)2 ImmutableMap (com.google.common.collect.ImmutableMap)1 File (java.io.File)1 IOException (java.io.IOException)1 AbstractMap (java.util.AbstractMap)1 ArrayList (java.util.ArrayList)1