I wrote:
> I noticed that avocet and trilobite (two of our CLOBBER_CACHE_ALWAYS
> animals) have started to fail on the deadlock-parallel isolation
> test, with symptoms that look like they're timing out.
> ...
> The simplest fix is to force that test to use debug_discard_caches=0,
> but I don't love that answer. Anybody have a better idea?
I thought of a better way: we can bypass the need to use a non-inlined
SQL function by declaring internal-language aliases for the
auxiliary-lock functions that are falsely marked parallel-safe.
It's a little bit ugly (see [1]) but I think it beats not running
this test under debug_discard_caches at all.
regards, tom lane
[1]
https://git.postgresql.org/gitweb/?p=postgresql.git;a=commitdiff;h=837cc73af29cd9c63515a6f2a36f54dd703a3a3f
