[
https://issues.apache.org/jira/browse/CALCITE-1052?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15095098#comment-15095098
]
Josh Elser commented on CALCITE-1052:
-------------------------------------
I've hacked together a little script locally based on what you said here:
{noformat}
==> create_integer_table.sql <==
drop table if exists integer_table;
create table integer_table (id integer not null primary key, column1 integer);
UPSERT INTO INTEGER_TABLE VALUES(1, 1);
UPSERT INTO INTEGER_TABLE VALUES(2, 2);
UPSERT INTO INTEGER_TABLE VALUES(3, 3);
==> create_string_table.sql <==
drop table if exists varchar_table;
create table varchar_table (id integer not null primary key, column1
varchar(50));
UPSERT INTO VARCHAR_TABLE VALUES(1, 'str_1');
UPSERT INTO VARCHAR_TABLE VALUES(2, 'str_2');
UPSERT INTO VARCHAR_TABLE VALUES(3, 'str_3');
{noformat}
I inserted 50k records this way (upserts 4 through 49999 omitted for obv
reasons).
I then ran two loops querying for records in the tables via sqlline-thin.py:
{noformat}
==> query_integer.sql <==
SELECT column1 FROM INTEGER_TABLE where id = 5000;
SELECT column1 FROM INTEGER_TABLE where id = 5000;
SELECT column1 FROM INTEGER_TABLE where id = 5000;
SELECT column1 FROM INTEGER_TABLE where id = 5000;
SELECT column1 FROM INTEGER_TABLE where id = 5000;
...
==> query_string.sql <==
SELECT column1 FROM VARCHAR_TABLE where id = 5000;
SELECT column1 FROM VARCHAR_TABLE where id = 5000;
SELECT column1 FROM VARCHAR_TABLE where id = 5000;
SELECT column1 FROM VARCHAR_TABLE where id = 5000;
SELECT column1 FROM VARCHAR_TABLE where id = 5000;
...
{noformat}
{noformat}
while true; do /usr/local/lib/phoenix/bin/sqlline-thin.py ~/query_string.sql;
done
{noformat}
{noformat}
while true; do /usr/local/lib/phoenix/bin/sqlline-thin.py ~/query_integer.sql;
done
{noformat}
Given what you described, [~kliew], I should expected to see "5000" showing up
while querying VARCHAR_TABLE and "str_5000" while querying INTEGER_TABLE (or
some errors casting "str_5000" into an integer). However, I did not see this.
I'll try updating Phoenix and Calcite to their most recent snapshots and
re-running to check for a regression, but I am not anticipating seeing what you
describe.
If you can create a test case which shows this actually happening, that would
extremely helpful. Right now, I can only assume that for your own driver, you
did not create unique connectionIDs. I'm still baffled about seeing this with
the JDBC driver as you said though.
> Phoenix queryserver sends result set in response to the wrong request when
> there are concurrent requests
> --------------------------------------------------------------------------------------------------------
>
> Key: CALCITE-1052
> URL: https://issues.apache.org/jira/browse/CALCITE-1052
> Project: Calcite
> Issue Type: Bug
> Reporter: Kevin Liew
> Assignee: Josh Elser
> Priority: Critical
> Labels: phoenix, queryserver
>
> Create two tables
> {code:sql}
> DROP TABLE IF EXISTS SEN.VARCHAR_TABLE;
> CREATE TABLE IF NOT EXISTS SEN.VARCHAR_TABLE(
> KeyColumn VARCHAR(255) PRIMARY KEY,
> Column1 VARCHAR(510));
> UPSERT INTO SEN.VARCHAR_TABLE VALUES ("One","1");
> {code}
> {code:sql}
> DROP TABLE IF EXISTS SEN.INTEGER_TABLE;
> CREATE TABLE IF NOT EXISTS SEN.INTEGER_TABLE(
> KeyColumn VARCHAR(255) PRIMARY KEY,
> Column1 INTEGER);
> UPSERT INTO SEN.VARCHAR_TABLE VALUES ("Two",2);
> {code}
> Running these two programs results in several crashes.
> 1. select a varchar by parameterized statement resulting in
> SELECT Column1 FROM SEN.VARCHAR_TABLE WHERE KeyColumn = 'One'
> {code:java}
> import java.sql.Connection;
> import java.sql.DriverManager;
> import java.sql.ResultSet;
> import java.sql.SQLException;
> import java.sql.PreparedStatement;
> import java.sql.Statement;
> public class Hello_World {
> public static void main(String[] args) throws SQLException {
> try {
> Class.forName("org.apache.phoenix.queryserver.client.Driver");
> } catch (ClassNotFoundException e) {
> System.out.println("Where is your PhoenixDriver");
> e.printStackTrace();
> return;
> }
> Connection conn =
> DriverManager.getConnection("jdbc:phoenix:thin:url=http://192.168.222.52:8765";);
> conn.setAutoCommit(true);
> String sqlStmt = "SELECT Column1 FROM SEN.VARCHAR_TABLE WHERE
> KeyColumn = ?";
> System.out.println("SQL Statement:\n\t" + sqlStmt);
>
> while(true)
> {
> ResultSet rset = null;
>
> //Statement stmt = conn.createStatement();
> PreparedStatement stmt = conn.prepareStatement(sqlStmt);
> stmt.setString(1, "One");
> ResultSet rs = stmt.executeQuery();
>
> while (rs.next()) {
> String column1 = rs.getString("column1");
> if (!column1.equals("1"))
> {
> System.out.println(column1);
> }
> }
> }
>
> //conn.close();
> }
>
> }
> {code}
> 2. select an integer by parameterized statement resulting in
> SELECT Column1 FROM SEN.INTEGER_TABLE WHERE KeyColumn = 'Two'
> {code:java}
> import java.sql.Connection;
> import java.sql.DriverManager;
> import java.sql.ResultSet;
> import java.sql.SQLException;
> import java.sql.PreparedStatement;
> import java.sql.Statement;
> public class Hello_World {
> public static void main(String[] args) throws SQLException {
> try {
> Class.forName("org.apache.phoenix.queryserver.client.Driver");
> } catch (ClassNotFoundException e) {
> System.out.println("Where is your PhoenixDriver");
> e.printStackTrace();
> return;
> }
> Connection conn =
> DriverManager.getConnection("jdbc:phoenix:thin:url=http://192.168.222.52:8765";);
> conn.setAutoCommit(true);
>
> String sqlStmt = "SELECT Column1 FROM SEN.INTEGER_TABLE WHERE
> KeyColumn = ?";
> System.out.println("SQL Statement:\n\t" + sqlStmt);
>
> while(true)
> {
> ResultSet rset = null;
>
> //Statement stmt = conn.createStatement();
> PreparedStatement stmt = conn.prepareStatement(sqlStmt);
> stmt.setString(1, "Two");
> ResultSet rs = stmt.executeQuery();
>
> while (rs.next()) {
> int column1 = rs.getInt("column1");
> if (column1 != 2)
> {
> System.out.println(column1);
> }
> }
> }
>
> //conn.close();
> }
>
> }
> {code}
> There are several crashes (might be preventable by adding a pause in the
> loops?), but the one relevant to this bug is:
> {code:java}
> SQL Statement:
> SELECT Column1 FROM SEN.INTEGER_TABLE WHERE KeyColumn = ?
> Exception in thread "main" java.lang.ClassCastException: java.lang.String
> cannot be cast to java.lang.Number
> at
> org.apache.calcite.avatica.util.AbstractCursor$NumberAccessor.getNumber(AbstractCursor.java:661)
> at
> org.apache.calcite.avatica.util.AbstractCursor$BigNumberAccessor.getInt(AbstractCursor.java:602)
> at
> org.apache.calcite.avatica.AvaticaResultSet.getInt(AvaticaResultSet.java:314)
> at Hello_World.main(Hello_World.java:36)
> {code}
> where we get a string from SEN.VARCHAR_TABLE while we are querying from the
> SEN.INTEGER_TABLE.
> The queryserver is sending the result set in response to a request made from
> another connection id. The statement id was not checked but there may have
> been a statement id collision
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
