Yes, the keys should be sorted, before passing them to HFile.Writer. The
way we build index for HFile is based on this assumption, that the keys are
sorted.
Mike Thomsen <mikerthom...@gmail.com> 于2019年1月1日周二 下午9:36写道：
> I took some of the code from the JUnit test for setting up a HFile writer
> and it looked like it should work.
>
> Path f = new Path("/", "test-something.hfile");
> HFileContext context = new HFileContextBuilder()
> .withBlockSize(4096)
> .withIncludesTags(true)
> .withCompression(Compression.Algorithm.NONE).build();
> HFile.Writer w = HFile.getWriterFactory(conf, cacheConf)
> .withPath(FileSystem.get(conf), f)
> .withFileContext(context)
> .withComparator(KeyValue.COMPARATOR)
> .create();
> //KeyValue(byte[] row, byte[] family, byte[] qualifier, long timestamp,
> byte[] value)
> Map<String, KeyValue> records = new HashMap<String, KeyValue>();
> for (int x = 0; x < 10000; x++) {
> String uuid = UUID.randomUUID().toString();
> KeyValue value = new KeyValue(uuid.getBytes(), "some-fam".getBytes(),
> "test".getBytes(), System.currentTimeMillis(), "hi".getBytes());
> w.append(value);
> }
>
> That threw this exception:
>
> java.io.IOException: Added a key not lexically larger than previous.
> Current cell =
>
> 8faf82ba-5fed-4332-9731-0ebf4d4494f9/some-fam:test/1546348994160/Put/vlen=2/seqid=0,
> lastCell =
>
> e32ec727-d946-4f67-a3c6-315a27c76408/some-fam:test/1546348994160/Put/vlen=2/seqid=0
> at
> org.apache.hadoop.hbase.io
> .hfile.AbstractHFileWriter.checkKey(AbstractHFileWriter.java:204)
> at
> org.apache.hadoop.hbase.io
> .hfile.HFileWriterV2.append(HFileWriterV2.java:267)
> at
> org.apache.hadoop.hbase.io
> .hfile.HFileWriterV3.append(HFileWriterV3.java:87)
> at HFileTest.main(HFileTest.java:39)
> at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
> at
>
> sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
> at
>
> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
> at java.lang.reflect.Method.invoke(Method.java:498)
> at org.codehaus.mojo.exec.ExecJavaMojo$1.run(ExecJavaMojo.java:282)
> at java.lang.Thread.run(Thread.java:748)
>
> When I manually sorted the keys, it worked fine for creating and closing
> the new HFile. Did I miss something or is this expected behavior from the
> writer? I'm testing this against HBase 1.2.9.
>
> Thanks,
>
> Mike
>
