Re:BUG #17533: run benchmarksql test failed

Hi,
Does anyone know why the GUC vacuum_defer_cleanup_age will affect the test result of benchmarkSQL on pgsql 14 ?
In addition, the tuple reported not found is visible in next query, i.e. the tuple does not lost.
And in pgsql 10, there never occur this issue.

Best regards.
At 2022-06-25 23:38:14, "PG Bug reporting form" <noreply(at)postgresql(dot)org> wrote:
>The following bug has been logged on the website:
>
>Bug reference: 17533
>Logged by: lx zou
>Email address: zoulx1982(at)163(dot)com
>PostgreSQL version: 14.2
>Operating system: Linux
>Description:
>
>Hi,
>
>Recently, i run test using benchmarksql on pg14, and sometimes my test
>will
>failed, and i want to know whether it's a bug.
>
>My test step is:
>1. download pg14 source code from here
>https://www.postgresql.org/ftp/source/v14.2/
>
>2. my operating system is linux centos7, here is my build step:
>./configure --prefix=`pwd`/inst --enable-debug CFLAGS="-O0"
>make -j > x
>make install
>./initdb -A trust --no-locale -E utf8 -D ../data
>
>postgres=# select version();
> version
>
>-----------------------------------------------------------------------------
> PostgreSQL 14.2 on x86_64-pc-linux-gnu, compiled by gcc (GCC) 9.1.0,
>64-bit
>(1 row)
>
>3. following is my postgresql.conf, others use default value
>
>vacuum_defer_cleanup_age = 60000
>listen_addresses = '*'
>port = 54321
>shared_buffers = 2GB
>
>4. the test tool i use is benchmarksql, which can download from here:
>https://sourceforge.net/projects/benchmarksql/
>and i use the lastet version benchmarksql-5.0.zip
>
>5. the test parameters is:
>warehouses=10
>loadWorkers=4
>terminals=10
>runTxnsPerTerminal=0
>runMins=150000
>limitTxnsPerMin=10000000
>terminalWarehouseFixed=false
>newOrderWeight=45
>paymentWeight=43
>orderStatusWeight=4
>deliveryWeight=4
>stockLevelWeight=4
>resultDirectory=my_result_%tY-%tm-%td_%tH%tM%tS
>osCollectorScript=./misc/os_collector_linux.py
>osCollectorInterval=1
>osCollectorDevices=net_eth0 blk_sda
>
>6. load data:
>./runSQL.sh props.pg ./sql.common/tableCreates.sql
>./runLoader.sh props.pg
>./runSQL.sh props.pg ./sql.common/indexCreates.sql
>./runSQL.sh props.pg ./sql.common/foreignKeys.sql
>./runSQL.sh props.pg ./sql.postgres/extraHistID.sql
>./runSQL.sh props.pg ./sql.postgres/buildFinish.sql
>
>7. run test:
>./runBenchmark.sh props.pg
>
>when run step 7, i offen receive such error:
><<<
>Term-00, Running Average tpmTOTAL: 95259.54 Current tpmTOTAL: 19452
>Memory Usage: 308MB / 1963MB
>19:56:14,868 [Thread-7] FATAL jTPCCTerminal : STOCK with S_W_ID=5
>S_I_ID=36110 not fount
>java.lang.Exception: STOCK with S_W_ID=5 S_I_ID=36110 not fount
> at jTPCCTData.executeNewOrder(jTPCCTData.java:483)
>>>>
>
>and i also test with different guc config, i found if do not set
>vacuum_defer_cleanup_age to 60000
>(i.e. use default value 0), the step 7 can run successfully.
>
>I also test pgsql 10.21 version, and no matter i set
>vacuum_defer_cleanup_age to 60000
>or use default value 0, the step 7 can always run ok.
>
>So, i sumerrize following results:
>
>14.2 version
>vacuum_defer_cleanup_age = 60000 failed
>vacuum_defer_cleanup_age = 0 ok
>10.21 version
>vacuum_defer_cleanup_age = 60000 ok
>vacuum_defer_cleanup_age = 0 ok
>
>In addition, i do not run test on pgsql 11 - 13 version, so i do not known
>their test results.
>
>Thank you for your time.
>
>Regards.
>