On Fri, Feb 9, 2018 at 1:36 AM, Masahiko Sawada <sawada.mshk@gmail.com> wrote:
> On Fri, Feb 9, 2018 at 12:45 AM, Claudio Freire <klaussfreire@gmail.com> wrote:
>> On Thu, Feb 8, 2018 at 1:36 AM, Masahiko Sawada <sawada.mshk@gmail.com> wrote:
>>> On Tue, Feb 6, 2018 at 9:51 PM, Claudio Freire <klaussfreire@gmail.com> wrote:
>>>> I can look into doing 3, that *might* get rid of the need to do that
>>>> initial FSM vacuum, but all other intermediate vacuums are still
>>>> needed.
>>>
>>> Understood. So how about that this patch focuses only make FSM vacuum
>>> more frequently and leaves the initial FSM vacuum and the handling
>>> cancellation cases? The rest can be a separate patch.
>>
>> Ok.
>>
>> Attached split patches. All but the initial FSM pass is in 1, the
>> initial FSM pass as in the original patch is in 2.
>>
>> I will post an alternative patch for 2 when/if I have one. In the
>> meanwhile, we can talk about 1.
>
> Thank you for updating the patch!
>
> + /* Tree pruning for partial vacuums */
> + if (threshold)
> + {
> + child_avail = fsm_get_avail(page, slot);
> + if (child_avail >= threshold)
> + continue;
> + }
>
> I think slots in a non-leaf page might not have a correct value
> because fsm_set_avail doesn't propagate up beyond pages. So do we need
> to check the root of child page rather than a slot in the non-lead
> page? I might be missing something though.
I'm not sure I understand what you're asking.
Yes, a partial FSM vacuum won't correct those inconsistencies. It
doesn't matter though, because as long as the root node (or whichever
inner node we're looking at the time) reports free space, the lookup
for free space will recurse and find the lower nodes, even if they
report more space than the inner node reported. The goal of making
free space discoverable will have been achieved nonetheless.
The final FSM vacuum pass isn't partial, to finally correct all those
small inconsistencies.