Thread: BUG #16699: PostgreSQL failed to run "timetz" test on release and debug configuration with MSVC on windows

The following bug has been logged on the website:

Bug reference:      16699
Logged by:          Gao Lin
Email address:      gaolin01@beyondsoft.com
PostgreSQL version: 13.0
Operating system:   Windows Server 2016
Description:

Environment:
Windows Server 2016 + VS2019 + PostgreSQL master branch on babef40 commit.

PostgreSQL failed to run "vcregress check parallel" due to one test "timetz"
on release and debug configuration with MSVC on windows. It can be produced
on master branch on babef40 commit. Could you please help look at this issue
?  Thanks a lot! 

And I have one question: How to disable a single test "timetz" in "vcregress
check parallel" command line? 

Repro steps:
1. git clone https://git.postgresql.org/git/postgresql.git
F:\gitP\git\postgresql.git
2. git -C "F:\gitP\git\postgresql.git" reset --hard babef40
3. cd F:\gitP\git\postgresql.git\src\tools\msvc
4. set
PATH=F:\gitP\git\postgresql.git\..\tools;F:\tools\gnuwin32\bin;%PATH%
5. build RELEASE
6. cd F:\gitP\git\postgresql.git
7. msbuild /m /p:Platform=x64 /p:Configuration=Release
/p:PlatformToolset=v141 pgsql.sln /verbosity:normal /t:Rebuild
8. cd F:\gitP\git\postgresql.git\..\install_amd64\bin
9. mkdir datadir
10. initdb.exe datadir
11. pg_ctl -D datadir -l logfile start -w -t 180
12. cd F:\gitP\git\postgresql.git\src\tools\msvc
13. vcregress check parallel

ErrorMessage:
parallel group (20 tests):  lseg numerology path line circle point time
timetz macaddr macaddr8 xid interval date inet tstypes strings box polygon
timestamp timestamptz
     strings                      ... ok          162 ms
     numerology                   ... ok           71 ms
     point                        ... ok           88 ms
     lseg                         ... ok           61 ms
     line                         ... ok           75 ms
     box                          ... ok          254 ms
     path                         ... ok           71 ms
     polygon                      ... ok          266 ms
     circle                       ... ok           72 ms
     date                         ... ok          123 ms
     time                         ... ok           79 ms
     timetz                       ... FAILED       82 ms
     timestamp                    ... ok          405 ms
     timestamptz                  ... ok          424 ms
     interval                     ... ok          109 ms
     inet                         ... ok          110 ms
     macaddr                      ... ok           79 ms
     macaddr8                     ... ok           86 ms
     tstypes                      ... ok          109 ms
     xid                          ... ok           94 ms

============== shutting down postmaster               ==============

========================
 1 of 200 tests failed. 
========================

The differences that caused some tests to fail can be viewed in the
file "F:/gitP/git/postgresql.git/src/test/regress/regression.diffs".  A copy
of the test summary that you see
above is saved in the file
"F:/gitP/git/postgresql.git/src/test/regress/regression.out".

Thanks,
Lin


On Tue, Nov 3, 2020 at 11:29 AM PG Bug reporting form
<noreply@postgresql.org> wrote:
>
> The following bug has been logged on the website:
>
> Bug reference:      16699
> Logged by:          Gao Lin
> Email address:      gaolin01@beyondsoft.com
> PostgreSQL version: 13.0
> Operating system:   Windows Server 2016
> Description:
>
> Environment:
> Windows Server 2016 + VS2019 + PostgreSQL master branch on babef40 commit.
>
> PostgreSQL failed to run "vcregress check parallel" due to one test "timetz"
> on release and debug configuration with MSVC on windows. It can be produced
> on master branch on babef40 commit. Could you please help look at this issue
> ?  Thanks a lot!

This is most likely the timetz issues that showed up with the US DST
changes, which have already been fixed in commit
4a071afbd056282746a5bc9362e87f579a56402d (along with backpatches). So
if you move to the latest version of your branch, it should go away.

That's a pretty stand-alone fix so you should be able to easily
backpatch it to earlier versions if needed.


> And I have one question: How to disable a single test "timetz" in "vcregress
> check parallel" command line?

The tests included in the parallel check are defined in
src/test/regress/parallel_schedule. You should be able to remove it
from there.

-- 
 Magnus Hagander
 Me: https://www.hagander.net/
 Work: https://www.redpill-linpro.com/