The first permutation done in the test does a wait, a wakeup then a
detach. It is proving to be unstable in the CI for FreeBSD (Windows and
Linux are stable). The failure shows that the wait is so slow to finish
after being woken up that the detach has the time to finish before the
wait, messing up with the expected output.
There may be a platform-specific issue going on here, but for now
disable this permutation to make the CI runs more stable.
Discussion: https://postgr.es/m/
[email protected] Parsed test spec with 2 sessions
-starting permutation: wait1 wakeup2 detach2
-injection_points_attach
------------------------
-
-(1 row)
-
-step wait1: SELECT injection_points_run('injection-points-wait'); <waiting ...>
-step wakeup2: SELECT injection_points_wakeup('injection-points-wait');
-injection_points_wakeup
------------------------
-
-(1 row)
-
-step wait1: <... completed>
-injection_points_run
---------------------
-
-(1 row)
-
-step detach2: SELECT injection_points_detach('injection-points-wait');
-injection_points_detach
------------------------
-
-(1 row)
-
-
starting permutation: wait1 detach2 wakeup2
injection_points_attach
-----------------------
step detach2 { SELECT injection_points_detach('injection-points-wait'); }
# Detach after wait and wakeup.
-permutation wait1 wakeup2 detach2
+# This permutation is proving to be unstable on FreeBSD, so disable for now.
+#permutation wait1 wakeup2 detach2
# Detach before wakeup. s1 waits until wakeup, ignores the detach.
permutation wait1 detach2 wakeup2