> 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 scans (one index scan here affects each and every index). Doing
> more than a single index scan is presumably rare, but are likely
> common among VACUUM operations that take an unusually long time --
> which is where the failsafe is relevant.
Ah, OK. I was confused by the terminology. I read "index scans" as a single
Index scan rather than a index scan cycle.
FWIW, even in the parallel case, the failsafe is checked after every index
scan cycle.
> Want to have a go at writing a patch for that?
Yes, I can.
Regards,
Sami Imseih
Amazon Web Services (AWS)