> I think as a first step, the deadlock detection should kick off after a
> certain timeout, even if the transaction timeout was not set.

> What do you think?

Dmitry, I thought that was what I suggested, no? =)

--Yakov

Reply via email to