Re: misc/57507: Automated test case slimit_le_1_user fails intermittently on some architectures

[Jim Spath <jspath55%gmail.com@localhost>]

The following reply was made to PR misc/57507; it has been noted by GNATS.

From: Jim Spath <jspath55%gmail.com@localhost>
To: gnats-bugs%netbsd.org@localhost
Cc: 
Subject: Re: misc/57507: Automated test case slimit_le_1_user fails
 intermittently on some architectures
Date: Wed, 12 Jul 2023 18:37:11 -0400

 On Fri, Jul 7, 2023 at 4:50=E2=80=AFAM Martin Husemann <martin%duskware.de@localhost>=
  wrote:
 >  Did the cron job run the single test you are reporting, or the full
 >  atf test suite?
 
 Thank you for the guidance. I had one cron job that ran a script with
 successive single ATF cases, such as:
 
 cd /usr/tests
 atf-run \
  fs/ffs/t_quotalimit:slimit_le_1_user
 atf-run \
  fs/vfs/t_full:nfs_fillfs
 [...]
 
 >  Were any processes left over (e.g. rump_server idle spinning) after the
 >  test run, and was there something making sure those were all killed
 >  after the run (or before the next one)?
 
 I looked back over one of the systems which had infrequent failures,
 and found these processes:
 
 root       4502  0.0  0.2  82596  6068 ?      Ssl  Thu12AM  18:13.56
 /usr/tests/fs/ffs/h_quota2_tests -b 1 fsimage unix://test
 root       5350  0.0  0.2  83332  6020 ?      Isl  Thu07PM  15:55.09
 /usr/tests/fs/ffs/h_quota2_tests -b 1 fsimage unix://test
 root      20843  0.0  0.2  82932  6040 ?      Isl  25Jun23  30:40.67
 /usr/tests/fs/ffs/h_quota2_tests -b 1 fsimage unix://test
 
 These times equate to the 3 failures on that system of this test case.
 
 info: atf.version, Automated Testing Framework 0.20 (atf-0.20)
 info: tests.root, /usr/tests
 info: time.start, Sun Jun 25 03:11:00 UTC 2023
 [...]
 tps-count: 1
 tp-start: 1687662660.702304, fs/ffs/t_quotalimit, 1
 tc-start: 1687662660.702379, slimit_le_1_user
 tc-so:Executing command [ newfs -q user -B le -O 1 -s 4000 -F fsimage ]
 tc-so:Executing command [ /usr/tests/fs/ffs/h_ffs_server -b fsimage
 unix://test ]
 tc-so:Executing command [ env LD_PRELOAD=3D/usr/lib/librumphijack.so
 RUMPHIJACK=3Dvfs=3Dgetvfsstat,blanket=3D/mnt edquota -u -s2k/4 -h3k/6 -t
 1s/1d 1 ]
 tc-so:Executing command [ rump.halt ]
 tc-so:Executing command [ /usr/tests/fs/ffs/h_quota2_tests -b 1
 fsimage unix://test ]
 tc-se:Fail: regexp test 1: write beyond the soft limit after grace
 time returned 69: Disc quota exceeded not in stderr
 tc-end: 1687662663.27629, slimit_le_1_user, failed, atf-check failed;
 see the output of the test for details
 tp-end: 1687662663.28139, fs/ffs/t_quotalimit
 info: time.end, Sun Jun 25 03:11:03 UTC 2023
 
 I did not have any steps in place to remove the orphan processes.
 
 I did not see any other unexplained processes on that system; I will
 review the others that are still "up". A couple of the smaller systems
 ran out of one resource or another and I've rebooted them since the
 test sweeps.
 
 Jim