Alvaro Herrera <alvherre@commandprompt.com> writes:
> Tom Lane wrote:
>> Actually, after thinking about this some more, I realize that this code
>> has got a significantly bigger problem than just whether it will respond
>> to CANCEL promptly.
> Err, that problem was exactly why I added the interrupt holdoff in
> there, so if you've got a better/more invasive solution, it's very
> welcome.
Well, that's a pretty incomplete solution :-(. Maybe we should do
something about this. There wasn't any obvious solution before,
but now that we have the nontransactional smgr-level sinval messages
being sent on drops and truncates, it seems like tying rd_targblock
clearing to those would fix the problem. The easiest way to do that
would involve moving rd_targblock down to the SMgrRelation struct.
Probably rd_fsm_nblocks and rd_vm_nblocks too. Comments?
regards, tom lane