[jira] [Created] (IGNITE-17041) Normalize query entity after it is modified during merge process.

["Mikhail Petrov (Jira)" <ji...@apache.org>]

Mikhail Petrov created IGNITE-17041:
---------------------------------------

             Summary: Normalize query entity after it is modified during merge process.
                 Key: IGNITE-17041
                 URL: https://issues.apache.org/jira/browse/IGNITE-17041
             Project: Ignite
          Issue Type: Bug
            Reporter: Mikhail Petrov


It is needed to to normalize query entity after it is modified during MERGE process as it is done during the first cache configuration processing. Currently new table columns that was created based on Query Entity fields which was added during MERGE process has the naming that differs from columns that were created based on initial Query Entity fields.

For example if CacheConfiguration#isSqlEscapeAll flag is disabled - all QueryEntity fields are converted to upper case and used as such to name columns. But it does not happen if Query Entity field was added during MERGE process. It confuses users and leads to the situations when column conflicts cannot be found because column names are different.

Reproducer:


{code:java}
public class TestClass extends GridCommonAbstractTest {
    /**
     * Start cluster nodes.
     */
    public static final int NODES_CNT = 2;

    /**
     * Count of backup partitions.
     */
    public static final int BACKUPS = 2;


    @Override
    protected IgniteConfiguration getConfiguration(String igniteInstanceName) throws Exception {
        QueryEntity queryEntity = new QueryEntity(String.class, Person.class)
                .setTableName("PERSON")
                .addQueryField("id", Boolean.class.getName(), null)
                .addQueryField("name", String.class.getName(), null);

        CacheConfiguration<?, ?> configuration = new CacheConfiguration<>(GridAbstractTest.DEFAULT_CACHE_NAME)
            .setBackups(BACKUPS)
            .setQueryEntities(Collections.singletonList(queryEntity));

        if (igniteInstanceName.endsWith("1"))
            queryEntity.addQueryField("age", Boolean.class.getName(), null);

        IgniteConfiguration cfg = super.getConfiguration(igniteInstanceName)
                .setConsistentId(igniteInstanceName)
                .setDataStorageConfiguration(new DataStorageConfiguration()
                        .setDefaultDataRegionConfiguration(new DataRegionConfiguration()))
                .setCacheConfiguration(
                    configuration);

        return cfg;
    }


    /**
     * {@inheritDoc}
     */
    @Override
    protected void afterTest() throws Exception {
        stopAllGrids();
    }

    /**
     *
     */
    @Test
    public void testIssue() throws Exception {
        startGrid(0);

        grid(0);

        grid(0).cache(GridAbstractTest.DEFAULT_CACHE_NAME).query(new SqlFieldsQuery("ALTER TABLE PERSON ADD age INTEGER")).getAll();

        GridTestUtils.assertThrows(log, () -> startGrid(1), Exception.class, "");

        grid(0).cluster().state(ClusterState.INACTIVE);

        startGrid(1);

        grid(0).cluster().state(ClusterState.ACTIVE);

        System.out.println(grid(0).cache(DEFAULT_CACHE_NAME).query(new SqlFieldsQuery("select * from \"SYS\".TABLE_COLUMNS"))
            .getAll());
    }

    class Person {
        private int id;

        private String name;

        private boolean age;
    }
}
{code}

As a result we can see that "age" column is duplicated in UPPER and CAMEL case. And no conflicts were found.
 



--
This message was sent by Atlassian Jira
(v8.20.7#820007)