Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

session, com_stmt: fetch all rows during EXECUTE command #42473

Merged
merged 9 commits into from
Mar 23, 2023

Conversation

YangKeao
Copy link
Member

@YangKeao YangKeao commented Mar 22, 2023

What problem does this PR solve?

Issue Number: close #42424, close #41891

What is changed and how it works?

  1. Fetch all rows and store them in the fetchedRows in the ResultSet.
  2. Close the result set in EXECUTE command, so we don't need to close it again in the following commands (FETCH, RESET and CLOSE)

Check List

Tests

  • Unit test
  • Integration test
  • Manual test (add detailed scripts or steps below)
  • No code

Test with the following a little bigger table:

import java.math.BigDecimal;
import java.sql.*;

public class Example {
    public static void main(String[] args) throws SQLException, InterruptedException, Exception {
        Connection conn = DriverManager.getConnection("jdbc:mysql://127.0.0.1:4000/test?useCursorFetch=true&useServerPrepStmts=true&useSSL=false", "root", "");

        conn.setAutoCommit(false); // must set

        conn.prepareStatement("drop table if exists t").execute();
        conn.prepareStatement("create table t (id int primary key, id_2 int)").execute();

        for (int i = 0; i < 10000; i++) {
            PreparedStatement stmt = conn.prepareStatement("insert into t values (?, ?)");
            stmt.setInt(1, i);
            stmt.setInt(2, i);
            stmt.execute();
        }

        // submit a statement with more arguments:
        PreparedStatement statement1 = conn.prepareStatement("" +
                "select * " +
                "from " +
                "t " +
                "where id > ? and id_2 > ?");
        statement1.setFetchSize(500);
        statement1.setBigDecimal(1, new BigDecimal(2));
        statement1.setBigDecimal(2, new BigDecimal(2));
        ResultSet rs_1 = statement1.executeQuery();

        // submit a statement with less arguments:
        PreparedStatement statement2 = conn.prepareStatement("" +
                "select * " +
                "from " +
                "t " +
                "where (id > ?)");
        statement2.setFetchSize(500);
        statement2.setBigDecimal(1, new BigDecimal(1));
        ResultSet rs_2 = statement2.executeQuery();


        // fetch the result from the first argument
        int iter = 3;
        while (rs_1.next()) {
            int result = rs_1.getInt(1);

            if (result != iter++ ) {
                throw new Exception("");
            }
        }
        System.out.println(iter);
        iter = 2;
        while (rs_2.next()) {
            int result = rs_2.getInt(1);

            if (result != iter++ ) {
                throw new Exception("");
            }
        }
        System.out.println(iter);

        conn.commit();
        statement1.close();
        statement2.close();
        conn.close();
    }
}

Side effects

  • Performance regression: Consumes more CPU
  • Performance regression: Consumes more Memory
  • Breaking backward compatibility

It stores all rows of a cursor fetch statement locally, while execuing the EXECUTE command, so it will eat up more memory (related with the size of query).

Release note

Previously, TiDB loads data in `FETCH` command with the `cursorFetch` enabled. Now, TiDB will load all data into the memory in `EXECUTE` command. It'll use more memory according to the size of result. And also, TiDB will never report write conflict in `FETCH` command with read-committed isolation level.

@YangKeao YangKeao requested a review from a team as a code owner March 22, 2023 08:03
@ti-chi-bot
Copy link
Member

ti-chi-bot commented Mar 22, 2023

[REVIEW NOTIFICATION]

This pull request has been approved by:

  • qw4990
  • tiancaiamao

To complete the pull request process, please ask the reviewers in the list to review by filling /cc @reviewer in the comment.
After your PR has acquired the required number of LGTMs, you can assign this pull request to the committer in the list by filling /assign @committer in the comment to help you merge this pull request.

The full list of commands accepted by this bot can be found here.

Reviewer can indicate their review by submitting an approval review.
Reviewer can cancel approval by submitting a request changes review.

@ti-chi-bot ti-chi-bot added needs-cherry-pick-release-7.0 release-note Denotes a PR that will be considered when it comes time to generate release notes. size/L Denotes a PR that changes 100-499 lines, ignoring generated files. labels Mar 22, 2023
@YangKeao YangKeao force-pushed the fix-stmt-executor-params branch 6 times, most recently from a71d076 to a231797 Compare March 22, 2023 08:16
@ti-chi-bot ti-chi-bot added size/XL Denotes a PR that changes 500-999 lines, ignoring generated files. and removed size/L Denotes a PR that changes 100-499 lines, ignoring generated files. labels Mar 22, 2023
@YangKeao
Copy link
Member Author

/retest

@zyguan
Copy link
Contributor

zyguan commented Mar 22, 2023

cc @jackysp , the other issue of cursor read, we need to tell users that it may cost more memory.

@ti-chi-bot ti-chi-bot added size/XXL Denotes a PR that changes 1000+ lines, ignoring generated files. and removed size/XL Denotes a PR that changes 500-999 lines, ignoring generated files. labels Mar 22, 2023
@ti-chi-bot ti-chi-bot added the status/LGT1 Indicates that a PR has LGTM 1. label Mar 22, 2023
@@ -524,7 +524,7 @@ func TestDispatchClientProtocol41(t *testing.T) {
com: mysql.ComStmtFetch,
in: []byte{0x1, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0},
err: nil,
out: []byte{0x5, 0x0, 0x0, 0x9, 0xfe, 0x0, 0x0, 0x82, 0x0},
out: []byte{0x5, 0x0, 0x0, 0x9, 0xfe, 0x0, 0x0, 0x42, 0x0},
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The original expected out is wrong. The fetchsize is 0 and the resultSet is not empty, so it shouldn't return 0x82 (which includes mysql.ServerStatusLastRowSend).
I modifed it to the correct answer here: the status is still ServerStatusCursorExists with no answer.

@YangKeao
Copy link
Member Author

/retest

@YangKeao
Copy link
Member Author

rebase the master. @hawkingrei seems to have made some optimization on test speed.

@hawkingrei
Copy link
Member

/merge

@ti-chi-bot
Copy link
Member

This pull request has been accepted and is ready to merge.

Commit hash: e613832

@ti-chi-bot ti-chi-bot added the status/can-merge Indicates a PR has been approved by a committer. label Mar 23, 2023
@YangKeao
Copy link
Member Author

/retest

1 similar comment
@YangKeao
Copy link
Member Author

/retest

@ti-chi-bot
Copy link
Member

In response to a cherrypick label: new pull request created to branch release-6.5: #42518.

ti-chi-bot pushed a commit to ti-chi-bot/tidb that referenced this pull request Mar 23, 2023
@ti-chi-bot
Copy link
Member

In response to a cherrypick label: new pull request created to branch release-7.0: #42519.

@tiancaiamao
Copy link
Contributor

close #38116

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
needs-cherry-pick-release-6.5 Should cherry pick this PR to release-6.5 branch. needs-cherry-pick-release-7.0 release-note Denotes a PR that will be considered when it comes time to generate release notes. size/XXL Denotes a PR that changes 1000+ lines, ignoring generated files. status/can-merge Indicates a PR has been approved by a committer. status/LGT2 Indicates that a PR has LGTM 2.
Projects
None yet
8 participants