Heikki Linnakangas <heikki.linnakangas@enterprisedb.com> writes:
> I still wonder, though, why we're seeing the error consistently on kudu,
> and not on any other animal. Perhaps the forknum field that's left
> uninitialized gets a different value there than on other platforms.
Hmm ... AFAICS this mistake would mean that no forknum field of the
requests[] array ever gets set at all, so they would stay at whatever
the virgin value in the shmem segment had been. Perhaps Solaris doesn't
guarantee that a shared memory block starts out as zeroes? But if
there were random garbage in the forknum fields you'd expect rather more
failures than are observed.
regards, tom lane