Andres Freund <and...@anarazel.de> writes: > I did that now. I couldn't reproduce it locally, despite a lot of > runs. Looking at the buildfarm it looks like the failures were, > excluding handfish which failed without recognizable symptoms before and > after, on BSD derived platforms (netbsd, freebsd, OX), which certainly > is interesting.
Isn't it now. Something about the BSD scheduler perhaps? But we've got four or five different BSD-ish platforms that reported failures, and it's hard to believe they've all got identical schedulers. That second handfish failure does match the symptoms elsewhere: https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=handfish&dt=2019-01-29%2000%3A20%3A22 --- /home/filiperosset/dev/client-code-REL_8/HEAD/pgsql.build/src/interfaces/ecpg/test/expected/thread-thread.stderr 2018-10-30 20:11:45.551967381 -0300 +++ /home/filiperosset/dev/client-code-REL_8/HEAD/pgsql.build/src/interfaces/ecpg/test/results/thread-thread.stderr 2019-01-28 22:38:20.614211568 -0200 @@ -0,0 +1,20 @@ +SQL error: page 0 of relation "test_thread" should be empty but is not on line 125 so it's not quite 100% BSD, but certainly the failure rate on BSD is way higher than elsewhere. Puzzling. regards, tom lane
