>I adjusted the FAILSAFE_EVERY_PAGES comments, which now point out that
>FAILSAFE_EVERY_PAGES is a power-of-two. The implication is that the
>compiler is all but guaranteed to be able to reduce the modulo
>division into a shift in the lazy_scan_heap loop, at the point of the
>
On Tue, Dec 20, 2022 at 9:44 AM Imseih (AWS), Sami wrote:
> Attached is a patch to check scanned pages rather
> than blockno.
Pushed, thanks!
I adjusted the FAILSAFE_EVERY_PAGES comments, which now point out that
FAILSAFE_EVERY_PAGES is a power-of-two. The implication is that the
compiler is
On Wed, Dec 21, 2022 at 2:44 AM Imseih (AWS), Sami wrote:
>
> Attached is a patch to check scanned pages rather
> than blockno.
Thank you for the patch. It looks good to me.
Regards,
--
Masahiko Sawada
Amazon Web Services: https://aws.amazon.com
On Sat, Nov 12, 2022 at 12:28 AM Imseih (AWS), Sami wrote:
>
> >Yeah, it's a little inconsistent.
>
> Yes, this should be corrected by calling the failsafe
> inside the parallel vacuum loops and handling the case by exiting
> the loop and parallel vacuum if failsafe kicks in.
I agree it's
Attached is a patch to check scanned pages rather
than blockno.
Regards,
Sami Imseih
Amazon Web Services (AWS)
v1-0001-fixed-when-wraparound-failsafe-is-checked.patch
Description: v1-0001-fixed-when-wraparound-failsafe-is-checked.patch
>Yeah, it's a little inconsistent.
Yes, this should be corrected by calling the failsafe
inside the parallel vacuum loops and handling the case by exiting
the loop and parallel vacuum if failsafe kicks in.
>I meant that there should definitely be a check between each round of
>index
On Thu, Nov 10, 2022 at 10:20 AM Imseih (AWS), Sami wrote:
> Consistency is the key point here. It is odd that a serial
> vacuum may skip the remainder of the indexes if failsafe
> kicks-in, but in the parallel case it will go through the entire index
> cycle.
Yeah, it's a little inconsistent.
>It makes sense to prefer consistency here, I suppose. The reason why
>we're not consistent is because it was easier not to be, which isn't
>exactly the best reason (nor the worst).
Consistency is the key point here. It is odd that a serial
vacuum may skip the remainder of the indexes
On Wed, Nov 9, 2022 at 6:29 AM Imseih (AWS), Sami wrote:
> When a user is running a parallel vacuum and the vacuum is long running
>
> due to many large indexes, it would make sense to check for failsafe earlier.
It makes sense to prefer consistency here, I suppose. The reason why
we're not
While looking through vacuum code, I noticed that
unlike non-parallel vacuum, parallel vacuum only gets
a failsafe check after an entire index cycle completes.
In vacuumlazy.c, lazy_check_wraparound_failsafe is checked
after every index completes, while in parallel, it is checked
after an entire
10 matches
Mail list logo
