Re: pgsql: Delay commit status checks until freezing executes.

Hi,

On 2023-01-11 14:29:25 -0800, Peter Geoghegan wrote:
> On Sat, Jan 7, 2023 at 7:25 PM Andres Freund <andres(at)anarazel(dot)de> wrote:
> > Probably a good idea, although it doesn't neatly fit right now.
>
> I'll leave it for now.
>
> Attached is v2, which changes things based on your feedback. Would
> like to get this out of the way soon.

Makes sense. It's clearly an improvement.

> + * We can't use TransactionIdDidAbort here because it won't treat transactions
> + * that were in progress during a crash as aborted by now. We determine that
> + * transactions aborted/crashed through process of elimination instead.

s/by now//?

> * When using an MVCC snapshot, we rely on XidInMVCCSnapshot rather than
> * TransactionIdIsInProgress, but the logic is otherwise the same: do not
> diff --git a/src/backend/access/transam/transam.c b/src/backend/access/transam/transam.c
> index 3a28dcc43..7629904bb 100644
> --- a/src/backend/access/transam/transam.c
> +++ b/src/backend/access/transam/transam.c
> @@ -110,7 +110,8 @@ TransactionLogFetch(TransactionId transactionId)
> * transaction tree.
> *
> * See also TransactionIdIsInProgress, which once was in this module
> - * but now lives in procarray.c.
> + * but now lives in procarray.c, as well as comments at the top of
> + * heapam_visibility.c that explain how everything fits together.
> * ----------------------------------------------------------------
> */

+1

Greetings,

Andres Freund