[
https://issues.apache.org/jira/browse/PHOENIX-2925?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Rajeshbabu Chintaguntla updated PHOENIX-2925:
---------------------------------------------
Attachment: PHOENIX-2925_v2.patch
bq. Rather than have another boolean flag for useLogicalName, can we always use
the logical name instead? Why do we only *sometimes* need to use the logical
name?
Some times we might need to directly write to physical table from phoenix
client for operations like deleteRows/upsertselectin in that case we need to
have physical table name. But in this specific case we need to use logicaltable
name. That's why added the flag.
bq. It will create a new file for each unique rowkey that come to the reducer.
Let me think that we can do with it.
Fixed it in the patch. Now it will create hfile only when table name changed.
Thanks.
> CsvBulkloadTool not working properly if there are multiple local indexes to
> the same table(After PHOENIX-1973)
> --------------------------------------------------------------------------------------------------------------
>
> Key: PHOENIX-2925
> URL: https://issues.apache.org/jira/browse/PHOENIX-2925
> Project: Phoenix
> Issue Type: Bug
> Affects Versions: 4.7.0
> Reporter: Rajeshbabu Chintaguntla
> Assignee: Rajeshbabu Chintaguntla
> Fix For: 4.8.0
>
> Attachments: PHOENIX-2925.patch, PHOENIX-2925.patch,
> PHOENIX-2925_v2.patch
>
>
> When there are multiple local indexes then only for first index data is
> getting generated properly and other indexes doesn't have any data. Changing
> testImportWithLocalIndex test as below is failing. ping [~sergey.soldatov]?
> {noformat}
> @Test
> public void testImportWithLocalIndex() throws Exception {
> Statement stmt = conn.createStatement();
> stmt.execute("CREATE TABLE TABLE6 (ID INTEGER NOT NULL PRIMARY KEY, "
> +
> "FIRST_NAME VARCHAR, LAST_NAME VARCHAR) SPLIt ON (1,2)");
> String ddl = "CREATE LOCAL INDEX TABLE6_IDX ON TABLE6 "
> + " (FIRST_NAME ASC)";
> stmt.execute(ddl);
> ddl = "CREATE LOCAL INDEX TABLE6_IDX2 ON TABLE6 " + " (LAST_NAME
> ASC)";
> stmt.execute(ddl);
> FileSystem fs = FileSystem.get(getUtility().getConfiguration());
> FSDataOutputStream outputStream = fs.create(new
> Path("/tmp/input3.csv"));
> PrintWriter printWriter = new PrintWriter(outputStream);
> printWriter.println("1,FirstName 1,LastName 1");
> printWriter.println("2,FirstName 2,LastName 2");
> printWriter.close();
> CsvBulkLoadTool csvBulkLoadTool = new CsvBulkLoadTool();
> csvBulkLoadTool.setConf(getUtility().getConfiguration());
> int exitCode = csvBulkLoadTool.run(new String[] {
> "--input", "/tmp/input3.csv",
> "--table", "table6",
> "--zookeeper", zkQuorum});
> assertEquals(0, exitCode);
> ResultSet rs = stmt.executeQuery("SELECT id, FIRST_NAME FROM TABLE6
> where first_name='FirstName 2'");
> assertTrue(rs.next());
> assertEquals(2, rs.getInt(1));
> assertEquals("FirstName 2", rs.getString(2));
> rs = stmt.executeQuery("SELECT LAST_NAME FROM TABLE6 where
> last_name='LastName 1'");
> assertTrue(rs.next());
> assertEquals("LastName 1", rs.getString(1));
> rs.close();
> stmt.close();
> }
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
