Re: Unstable tests for recovery conflict handling

> On Apr 27, 2022, at 9:45 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> [ starting a new thread cuz the shared-stats one is way too long ]
>
> Andres Freund <andres@anarazel.de> writes:
>> Add minimal tests for recovery conflict handling.
>
> It's been kind of hidden by other buildfarm noise, but
> 031_recovery_conflict.pl is not as stable as it should be [1][2][3][4].
>
> Three of those failures look like

Interesting,

I have been getting failures on REL_14_STABLE:

t/012_subtransactions.pl ............. 11/12
#   Failed test 'Rollback of PGPROC_MAX_CACHED_SUBXIDS+ prepared transaction on promoted standby'
#   at t/012_subtransactions.pl line 211.
#          got: '3'
#     expected: '0'
t/012_subtransactions.pl ............. 12/12 # Looks like you failed 1 test of 12.
t/012_subtransactions.pl ............. Dubious, test returned 1 (wstat 256, 0x100)
Failed 1/12 subtests

And the logs, tmp_check/log/regress_log_012_subtransactions, showing:

### Enabling streaming replication for node "primary"
### Starting node "primary"
# Running: pg_ctl -D
/Users/mark.dilger/recovery_test/postgresql/src/test/recovery/tmp_check/t_012_subtransactions_primary_data/pgdata-l
/Users/mark.dilger/recovery_test/postgresql/src/test/recovery/tmp_check/log/012_subtransactions_primary.log-o
--cluster-name=primarystart 
waiting for server to start.... done
server started
# Postmaster PID for node "primary" is 46270
psql:<stdin>:1: ERROR:  prepared transaction with identifier "xact_012_1" does not exist
not ok 11 - Rollback of PGPROC_MAX_CACHED_SUBXIDS+ prepared transaction on promoted standby

#   Failed test 'Rollback of PGPROC_MAX_CACHED_SUBXIDS+ prepared transaction on promoted standby'
#   at t/012_subtransactions.pl line 211.
#          got: '3'
#     expected: '0'


This is quite consistent for me, but only when I configure with --enable-coverage and --enable-dtrace.  (I haven't yet
triedone of those without the other.) 

I wasn't going to report this yet, having not yet completely narrowed this down, but I wonder if anybody else is seeing
this?

I'll try again on master....

—
Mark Dilger
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company