Bug #132799 for DBD-Pg: t/09arrays.t fails sometimes

Wed Jun 10 07:24:51 2020 TINITA [...] cpan.org - Ticket created

Subject:

t/09arrays.t fails sometimes

We are seeing test failures in Open Build Service (OBS) for t/09arrays.t, but not always. Here is a failure for x86_64 (version 3.12.1): https://build.opensuse.org/package/live_build_log/devel:languages:perl/perl-DBD-Pg/openSUSE_Leap_15.1/x86_64 But for other openSUSE versions it passes: https://build.opensuse.org/package/show/devel:languages:perl/perl-DBD-Pg However, on ppc64, I can see the test failure almost every time, with 3.12.3 https://build.opensuse.org/package/live_build_log/devel:languages:perl:autoupdate/perl-DBD-Pg/standard/ppc64 [ 96s] t/09arrays.t ........... [ 96s] All 203 subtests passed [ 96s] t/10_pg_error_field.t .. ok [ 97s] t/12placeholders.t ..... ok [ 97s] t/20savepoints.t ....... ok [ 98s] t/30unicode.t .......... ok [ 98s] t/99cleanup.t .......... ok [ 98s] [ 98s] Test Summary Report [ 98s] ------------------- [ 98s] t/09arrays.t (Wstat: 11 Tests: 203 Failed: 0) [ 98s] Non-zero wait status: 11 [ 98s] Files=17, Tests=2880, 32 wallclock secs ( 0.62 usr 0.05 sys + 8.71 cusr 1.07 csys = 10.45 CPU) [ 98s] Result: FAIL [ 98s] Failed 1/17 test programs. 0/2880 subtests failed. [ 98s] make: *** [Makefile:988: test_dynamic] Error 255 [ 98s] error: Bad exit status from /var/tmp/rpm-tmp.7c1Fqw (%check) Maybe the cleanup code has problems (as the tests itself are passing). I can test possible fixes in OBS if necessary.

Wed Jun 10 11:54:24 2020 TINITA [...] cpan.org - Correspondence added

Another observation I made is: When running tests with TEST_HARNESS=1 make test everything is fine. Also I can only reproduce the failure in OBS with version 3.12.3, not with 3.12.2 or 3.12.1

Wed Jun 10 12:30:26 2020 greg [...] turnstep.com - Correspondence added

Thank you for the report. Could you help us to debug it further by running that test in verbose mode? Should be doable from command line like so: prove -b -v t/09_arrays.t

Wed Jun 10 12:30:27 2020 The RT System itself - Status changed from 'new' to 'open'

Wed Jun 10 12:30:28 2020 greg [...] turnstep.com - Taken

Wed Jun 10 12:37:25 2020 TINITA [...] cpan.org - Correspondence added

On Wed Jun 10 18:30:26 2020, TURNSTEP wrote: Show quoted text

> Thank you for the report. Could you help us to debug it further by > running that test in verbose mode?

Yeah, I tried that already, see above. With the result that it always passes. That's also the workaround we are using now ;-)

Wed Jun 10 12:39:29 2020 TINITA [...] cpan.org - Correspondence added

On Wed Jun 10 18:37:25 2020, TINITA wrote: Show quoted text

> Yeah, I tried that already, see above. > With the result that it always passes. > That's also the workaround we are using now ;-)

Oh wait, sorry, I made a typo in my previous message, I meant: HARNESS_VERBOSE=1 make test

Wed Jun 10 12:47:42 2020 TINITA [...] cpan.org - Correspondence added

On Wed Jun 10 18:39:29 2020, TINITA wrote: Show quoted text

> Oh wait, sorry, I made a typo in my previous message, I meant: > > HARNESS_VERBOSE=1 make test

actually that doesn't result in verbose tests. I tried `prove -v` now, but it has the same effect, tests are passing.

Wed Jun 10 12:54:11 2020 TINITA [...] cpan.org - Correspondence added

I attached the log from 09arrays.t

Subject:

09arrays.log

Message body is not shown because it is too large.

Wed Jun 10 14:59:33 2020 greg [...] turnstep.com - Correspondence added

Hmm...this is hard to debug from a distance, especially with no verbose mode! :) Perhaps after running that single test, you could check for any database errors. The file should be dbdpg_test_database/data/pg_log/postgres2020-06.10.log Could also try adding to the bottom of t/09arrays.t: pass("Cleanup complete");

Tue Jul 14 10:52:54 2020 greg [...] turnstep.com - Correspondence added

I spun up an openSUSE box on AWS, but was unable to duplicate the error. Is there a test box I could have access to that will reliably fail the test?