Re: [GENERAL] Slow PITR restore

Hello Hannu,

Hannu Krosing wrote:
>> (For parallelized queries, superuser privileges might appear wrong, but
>> I'm arguing that parallelizing the rights checking isn't worth the
>> trouble, so the initiating worker backend should do that and only
>> delegate safe jobs to hepler backends. Or is that a serious limitation
>> in a way?)
>
> at least functions defined with SECURITY DEFINER; may be a problem

Uhm.. what I had in mind was parallelizing seqential scans, index scans,
joins and such - database internal stuff.

Parallelizing user defined functions (or what did you have in mind?) is
more difficult and sometimes impossible, because the planner cannot know
ahead, what the function's going to do.

However, thinking about it, maybe, one could and should try to
parallelize computationally intensive IMMUTABLE functions. But already
with STABLE ones I'm getting suspicious. It would require users to write
real thread-safe (well, multi-process-safe) functions, which I doubt
somewhat. Granted, they theoretically *should* be safe, but...

Anyway, if that's the only show stopper, one could certainly tell helper
backends to substitute their superuser privileges with the invoker's
privileges. Not sure if that's worth the trouble, though.

Regards

Markus