Hikari pool implementation in perfservice

[shaharuk-yb]

Flag based hikari pool support in benchbase.

Use flag use_hikari_pool: true/false in input yaml or xml. Default value is false

Optionally, you can set max pool size for hikari through property max_pool_size: <pool_size>. Default max pool size = terminals

Example yaml:

type: YUGABYTE
driver: com.yugabyte.Driver
url: jdbc:yugabytedb://localhost:5433/yugabyte?sslmode=disable&ApplicationName=featurebench&reWriteBatchedInserts=true&load-balance=false
username: yugabyte
password:
use_hikari_pool: true
max_pool_size: 10

batchsize: 128
isolation: TRANSACTION_REPEATABLE_READ
loaderthreads: 1
terminals: 1
collect_pg_stat_statements: true
use_dist_in_explain : true

works:
    work:
        time_secs: 5
        active_terminals: 1
        rate: unlimited
        warmup: 0
microbenchmark:
    class: com.oltpbenchmark.benchmarks.featurebench.customworkload.YBDefaultMicroBenchmark
    properties:
        setAutoCommit: true
        create:
            - DROP TABLE IF EXISTS pkey_rangescan_for_INclause_tbl_1;
            - CREATE TABLE pkey_rangescan_for_INclause_tbl_1(col_bigint_id_1 bigint, col_bigint_1 bigint, col_bigint_2 bigint, col_bigint_3 bigint, col_bigint_4 bigint, col_bigint_5 bigint, col_float2_1 float(2), col_float2_2 float(2), col_float5_1 float(5), col_float5_2 float(5), col_boolean_1 boolean, col_varchar10_1 varchar(10), col_varchar100_1 varchar(100), col_varchar100_2 varchar(100), col_varchar500_1 varchar(500), PRIMARY KEY(col_bigint_id_1 ASC));
        cleanup:
            - DROP TABLE IF EXISTS pkey_rangescan_for_INclause_tbl_1;
        loadRules:
            - table: pkey_rangescan_for_INclause_tbl_
              count: 1
              rows: 1000
              columns:
                  - name: col_bigint_id_
                    count: 1
                    util: PrimaryIntGen
                    params: [1, 1000]
                  - name: col_bigint_
                    count: 5
                    util: RandomNumber
                    params: [1, 1000]
                  - name: col_float2_
                    count: 2
                    util: RandomNoWithDecimalPoints
                    params: [1, 1000, 2]
                  - name: col_float5_
                    count: 2
                    util: RandomNoWithDecimalPoints
                    params: [1, 1000, 5]
                  - name: col_boolean_
                    count: 1
                    util: RandomBoolean
                  - name: col_varchar10_
                    count: 1
                    util: RandomAString
                    params: [10, 10]
                  - name: col_varchar100_
                    count: 2
                    util: RandomAString
                    params: [100, 100]
                  - name: col_varchar500_
                    count: 1
                    util: RandomAString
                    params: [500, 500]
        executeRules:
            - workload: scanG4_pkey_INclause_rangescan_increasing_rows_1row
              run:
                  - name: pkey_INclause_rangescan_increasing_rows_1row_test
                    weight: 100
                    queries:
                        - query: select col_bigint_id_1 from pkey_rangescan_for_INclause_tbl_1 where col_bigint_id_1 IN (?)
                          bindings:
                              count: 1
                              util: RandomNumber
                              params: [ 1, 1000 ]

[hbhanawat]

Why do we have to create a connection and then close it? I think we shouldn't be doing anything in the teardown for hikari.

[shaharuk-yb]

done

[hbhanawat]

Since worker inherits from benchmarkmodule are we creating one hikari pool per worker?

[shaharuk-yb]

even though worker inherits from benchmarkmodule, the datasource is created in benchmarkmodule's constructor which is called only once. Hence, there is only one hikari pool.

[hbhanawat]

We should shutdown the dataSource and its associated pool after the test is done. You can call datasource.close for the same.

[shaharuk-yb]

implemented closeDataSource() method which will close it in ThreadBench finalizeWorkers()

[shaharuk-yb]

@hbhanawat - Getting following error after closedatasource implementation.

HikariDataSource HikariDataSource (HikariPool-1) has been closed.
	at com.zaxxer.hikari.HikariDataSource.getConnection(HikariDataSource.java:96)
	at com.oltpbenchmark.api.BenchmarkModule.makeConnection(BenchmarkModule.java:94)
	at com.oltpbenchmark.api.BenchmarkModule.refreshCatalog(BenchmarkModule.java:242)
	at com.oltpbenchmark.DBWorkload.main(DBWorkload.java:618)