On Wed, Feb 8, 2023 at 2:28 PM Andres Freund <and...@anarazel.de> wrote:
> 2023-02-08 00:53:20.257 GMT client backend[4584] pg_regress/rangetypes
> STATEMENT: select '-[a,z)'::textrange;
> TRAP: failed Assert("PMSignalState->PMChildFlags[slot] ==
> PM_CHILD_ASSIGNED"), File: "../src/backend/storage/ipc/pmsignal.c", Line:
> 329, PID: 5948
No idea what's going on yet, but this assertion failure is very
familiar to me, as one of the ways that lorikeet fails/failed (though
it hasn't failed like that since the postmaster latchification).
There it was because Cygwin's signal blocking is unreliable, so the
postmaster could start a backend, while already being in the middle of
starting a backend. That particular problem shouldn't be possible
anymore; now we can only start backends from inside the main event
loop. Hmm. (State machine bug? Some confusion about processes
caused by the fact that PID was recycled?)
