Add a new GitHub workflow check designed to execute JDBC tests in parallel query mode

.github/composite-actions/install-extensions/action.yml

@@ -9,6 +9,10 @@ inputs:
     description: 'Database migration mode'
     required: no
     default: "single-db"
+  parallel_query_mode:
+    description: 'Postgres Parallel Query Mode'
+    required: no
+    default: false
 
 runs:
   using: "composite"
@@ -26,6 +30,6 @@ runs:
         sudo echo "host    all             all             $ipaddress/32            trust" >> pg_hba.conf
         ~/${{inputs.install_dir}}/bin/pg_ctl -D ~/${{inputs.install_dir}}/data/ -l logfile restart
         cd ~/work/babelfish_extensions/babelfish_extensions/
-        sudo ~/${{inputs.install_dir}}/bin/psql -d postgres -U runner -v user="jdbc_user" -v db="jdbc_testdb" -v migration_mode=${{inputs.migration_mode}} -f .github/scripts/create_extension.sql
+        sudo ~/${{inputs.install_dir}}/bin/psql -d postgres -U runner -v user="jdbc_user" -v db="jdbc_testdb" -v migration_mode=${{inputs.migration_mode}} -v parallel_query_mode=${{inputs.parallel_query_mode}} -f .github/scripts/create_extension.sql
         sqlcmd -S localhost -U "jdbc_user" -P 12345678 -Q "SELECT @@version GO"
       shell: bash

.github/scripts/create_extension.sql

@@ -8,5 +8,15 @@ GRANT ALL ON SCHEMA sys to :user;
 ALTER USER :user CREATEDB;
 ALTER SYSTEM SET babelfishpg_tsql.database_name = :db;
 ALTER SYSTEM SET babelfishpg_tsql.migration_mode = :'migration_mode';
+
+\if :parallel_query_mode
+    ALTER SYSTEM SET parallel_setup_cost = 0;
+    ALTER SYSTEM SET parallel_tuple_cost = 0;
+    ALTER SYSTEM SET min_parallel_index_scan_size = 0;
+    ALTER SYSTEM SET min_parallel_table_scan_size = 0;
+    ALTER SYSTEM SET force_parallel_mode = 1;
+    ALTER SYSTEM SET max_parallel_workers_per_gather = 4;
+\endif
+
 SELECT pg_reload_conf();
 CALL SYS.INITIALIZE_BABELFISH(:'user');

.github/workflows/jdbc-tests-with-parallel-query.yml

@@ -0,0 +1,86 @@
+name: JDBC Tests With Parallel Query
+on: [push, pull_request]
+
+jobs:
+  run-babelfish-jdbc-tests-with-parallel-query-mode:
+    runs-on: ubuntu-20.04
+    steps:
+      - uses: actions/checkout@v2
+        id: checkout
+
+      - name: Install Dependencies
+        id: install-dependencies
+        if: always()
+        uses: ./.github/composite-actions/install-dependencies
+
+      - name: Build Modified Postgres
+        id: build-modified-postgres
+        if: always() && steps.install-dependencies.outcome == 'success'
+        uses: ./.github/composite-actions/build-modified-postgres
+
+      - name: Compile ANTLR
+        id: compile-antlr
+        if: always() && steps.build-modified-postgres.outcome == 'success'
+        uses: ./.github/composite-actions/compile-antlr
+
+      - name: Build Extensions
+        id: build-extensions
+        if: always() && steps.compile-antlr.outcome == 'success'
+        uses: ./.github/composite-actions/build-extensions
+
+      - name: Install Extensions
+        id: install-extensions
+        if: always() && steps.build-extensions.outcome == 'success'
+        uses: ./.github/composite-actions/install-extensions
+        with:
+          parallel_query_mode: true
+
+      - name: Run JDBC Tests
+        id: jdbc
+        if: always() && steps.install-extensions.outcome == 'success'
+        timeout-minutes: 60
+        run: |
+          cd test/JDBC/
+          # set env variable isParallelQueryMode to true to let jdbc know not to run tests of file parallel_query_jdbc_schedule
+          export isParallelQueryMode=true
+          mvn test
+          # reset env variable
+          unset isParallelQueryMode
+
+      - name: Cleanup babelfish database
+        id: cleanup
+        if: always() && steps.install-extensions.outcome == 'success'
+        run: |
+          sudo ~/postgres/bin/psql -d postgres -U runner -v user="jdbc_user" -v db="jdbc_testdb" -f .github/scripts/cleanup_babelfish_database.sql
+
+      - name: Upload Log
+        if: always() && steps.jdbc.outcome == 'failure'
+        uses: actions/upload-artifact@v2
+        with:
+          name: postgres-log
+          path: ~/psql/data/logfile
+
+      # The test summary files contain paths with ':' characters, which is not allowed with the upload-artifact actions
+      - name: Rename Test Summary Files
+        id: test-file-rename
+        if: always() && steps.jdbc.outcome == 'failure'
+        run: |
+          cd test/JDBC/Info
+          timestamp=`ls -Art | tail -n 1`
+          cd $timestamp
+          mv $timestamp.diff ../output-diff.diff
+          mv "$timestamp"_runSummary.log ../run-summary.log
+
+      - name: Upload Run Summary 
+        if: always() && steps.test-file-rename == 'success'
+        uses: actions/upload-artifact@v2
+        with:
+          name: run-summary.log
+          path: test/JDBC/Info/run-summary.log
+
+      - name: Upload Output Diff
+        if: always() && steps.jdbc.outcome == 'failure'
+        uses: actions/upload-artifact@v2
+        with:
+          name: output-diff.diff
+          path: test/JDBC/Info/output-diff.diff

test/JDBC/README.md

@@ -18,6 +18,7 @@ The JDBC test framework for Babelfish uses the JDBC Driver for SQL Server for da
   - [IMPORTANT](#important)
 - [Adding the test cases](#adding-the-test-cases)
 - [Reading the console output and diff](#reading-the-console-output-and-diff)
+- [Running Tests with Parallel Query Enabled](#running-tests-with-parallel-query-enabled)
 
 ## Running the test framework
 
@@ -382,3 +383,34 @@ You will also be provided with the location of the `.diff` file containing the d
 - `mm`   → minutes
 - `ss`   → seconds
 - `SSS`  → milliseconds
+
+
+## Running Tests with Parallel Query Enabled
+
+After building the modified PostgreSQL engine and Babelfish extensions using the [online instructions](../../contrib/README.md), you must:
+1. Create a PostgreSQL database and initialize Babelfish (if you already have a database with Babelfish initialized, you can omit this step or perform the cleanup steps before you initialize) to enable parallel query mode pass -enable_parallel_query flag when running ./init.sh
+
+   ```bash
+    ./init.sh -enable_parallel_query
+    ```
+3. Before running JDBC tests, please take note that currently not all JDBC tests runs sucessfully with parallel query mode on. Certain JDBC tests are encountering issues, such as crashes, failures, or timeouts when executed with parallel query mode enabled. So we need these problematic tests to be excluded from running via jdbc framework. File  `parallel_query_jdbc_schedule` contains test-cases names with prefix `ignore#!#` that are problematic and needs to be avoided from being run. To exclude these problematic tests from running via the JDBC framework, use the `isParallelQueryMode` environment variable. You can set it to `true`:
+
+   ```bash
+    export isParallelQueryMode=true
+    # Verify if isParallelQueryMode is set to true
+    echo $isParallelQueryMode
+   ```
+4. Now Run the tests:
+    ```bash
+    mvn test
+    ```
+5. Cleanup all the objects, users, roles and databases created while running the tests:
+    ```bash
+    ./cleanup.sh
+    ```
+6. Please note that when you have completed testing with parallel query mode enabled, you should unset the `isParallelQueryMode` environment variable that was previously set to `true`. This ensures that all tests run in the normal Babelfish mode (without parallel query):
+
+   ```bash
+    unset isParallelQueryMode
+   ```
+If you encounter failing or crashing tests in the "JDBC tests with parallel query" GitHub workflow, consider adding the names of these problematic test cases to the `parallel_query_jdbc_schedule` file. Prefix these test case names with `ignore#!#`. As we work towards resolving these issues in the future, we will gradually remove these excluded tests from the `parallel_query_jdbc_schedule` scheduling file.

test/JDBC/init.sh

@@ -1,5 +1,24 @@
-
-#create test user and database from psql terminal
+# Default values
+parallel_query_mode=false
+
+# Parse command-line flags/arguments
+while [[ $# -gt 0 ]]; do
+    key="$1"
+    case $key in
+        -enable_parallel_query)
+        parallel_query_mode=true
+        shift
+        ;;
+      *)
+        # Unknown option
+        exit 1
+        ;;
+    esac
+    shift
+done
+
+# create test user and database from psql terminal
+if [[ $parallel_query_mode = false ]]; then
 echo "============================== CREATING USER AND DATABASE =============================="
 psql -U "$USER" -d postgres -a << EOF
 CREATE USER jdbc_user WITH SUPERUSER CREATEDB CREATEROLE PASSWORD '12345678' INHERIT;
@@ -17,3 +36,31 @@ show babelfishpg_tsql.database_name;
 CALL sys.initialize_babelfish('jdbc_user');
 EOF
 echo "============================= BUILDING JDBC TEST FRAMEWORK ============================="
+fi
+
+# create test user and database from psql terminal when parallel query mode is on
+if [[ $parallel_query_mode = true ]]; then
+echo "============================== CREATING USER AND DATABASE =============================="
+psql -U "$USER" -d postgres -a << EOF
+CREATE USER jdbc_user WITH SUPERUSER CREATEDB CREATEROLE PASSWORD '12345678' INHERIT;
+DROP DATABASE IF EXISTS jdbc_testdb;
+CREATE DATABASE jdbc_testdb OWNER jdbc_user;
+\c jdbc_testdb
+CREATE EXTENSION IF NOT EXISTS "babelfishpg_tds" CASCADE;
+GRANT ALL ON SCHEMA sys to jdbc_user;
+ALTER USER jdbc_user CREATEDB;
+\c jdbc_testdb
+ALTER SYSTEM SET babelfishpg_tsql.database_name = 'jdbc_testdb';
+ALTER SYSTEM SET parallel_setup_cost = 0;
+ALTER SYSTEM SET parallel_tuple_cost = 0;
+ALTER SYSTEM SET min_parallel_index_scan_size = 0;
+ALTER SYSTEM SET min_parallel_table_scan_size = 0;
+ALTER SYSTEM SET force_parallel_mode = 1;
+ALTER SYSTEM SET max_parallel_workers_per_gather = 4;
+SELECT pg_reload_conf();
+\c jdbc_testdb
+show babelfishpg_tsql.database_name;
+CALL sys.initialize_babelfish('jdbc_user');
+EOF
+echo "============================= BUILDING JDBC TEST FRAMEWORK ============================="
+fi