"Steven Flatt" <steven(dot)flatt(at)gmail(dot)com> writes:
> (gdb) p *bufHdr
> $1 = {tag = {rnode = {spcNode = 1663, dbNode = 16384, relNode = 85707},
> blockNum = 0}, flags = 70, usage_count = 5, refcount = 2,
> wait_backend_pid = 35775, buf_hdr_lock = 0 '\0', buf_id = 14407,
> freeNext = -2, io_in_progress_lock = 28933, content_lock = 28934}
Hm, PIN_COUNT_WAITER flag is still set, and refcount = 2 saying there is
still someone else pinning the buffer, so nothing evidently wrong here.
Could you check PrivateRefCount[14407] in both cores?
regards, tom lane