On Thu, Oct 26, 2023 at 1:41 PM Andres Freund wrote:
> Does it really matter on that level for the user whether a snapshot exists
> because of repeatable read or because of a cursor? If users don't understand
> backend_xmin - likely largely true - then the consequences of holding a
> snapshot ope
Hi,
On 2023-10-26 11:47:32 -0400, Robert Haas wrote:
> I've seen situations a few times now where somebody has sessions that
> are "idle in transaction" for a long time but they feel like it should
> be harmless because the transaction has no XID. However, the fact that
> the transaction is idle d
Hi,
I've seen situations a few times now where somebody has sessions that
are "idle in transaction" for a long time but they feel like it should
be harmless because the transaction has no XID. However, the fact that
the transaction is idle doesn't mean it isn't running a query, because
there could
